ࡱ> WULMNOPQX5@ bjbj22 \XXlbbbb~~~$. . . P~ L lB P"c>|B@@@$}Rd~)_c))dbby(D(D(D)b~(D)(D(DBE|X&~6  . ";@`00y0@ybbbby~!(D#_%+ddd?YdXjWD(Yj MSBNx: A Component-Centric Toolkit for Modeling and Inference with Bayesian Networks Carl M. Kadie David Hovel Eric Horvitz 28 July 2001 Technical Report MSR-TR-2001-67 Microsoft Research Microsoft Corporation One Microsoft Way Redmond, WA 98052 MSBNx: A Component-Centric Toolkit for Modeling and Inference with Bayesian Networks Carl M. Kadie David Hovel Eric Horvitz Adaptive Systems and Interaction Microsoft Research One Microsoft Way Redmond, WA 98052-6399 {carlk, davidhov, horvitz}@microsoft.com Abstract We review the functionality of a modular, component-based tool kit for Bayesian network development and inference. Beyond its operation as a standalone modeling and inference environment, MSBNx facilitates the development of standalone applications by providing a set of run-time components that provide Bayesian reasoning services when integrated into other programs. MSBNx inferential operations provide both inference about states of inference and about the value of information for unobservered evidence. The services and modeling environment supports both diagnostic and troubleshooting mingles observations and repair operations. MSBNx facilitates the development and use of new add-in components. The modeling environment provides a means for assessing distinctions and beliefs, and special interfaces and tools for representing the asymmetric nature of probability distributions. Introduction MSBNx is a component-based Windows application for creating, assessing, and evaluating Bayesian networks. MSBNx can provide diagnosis as well as troubleshooting inference that considers both observations and repair operations. Components of MSBNx can be integrated into programs allow them to leverage inference and decision making under uncertainty. When doing diagnosis and troubleshooting, MSBNx can recommend what evidence to gather next based on value of information (VOI) computations. Beyond using the components that MSBNx provides, developers and researchers can create their own new add-in components and use them within MSBNx. For information on downloading MSBNx and restrictions on its use, see  HYPERLINK "http://research.microsoft.com/adapt/MSBNx/" http://research.microsoft.com/adapt/MSBNx/. The application's installation module includes complete help files and sample networks. Bayesian networks are encoded in an XML-based file format. The application and its components run on any Windows system version from Windows 98 to Windows XP. MSBNx tries to make it easy for you to specify your probabilities for a Bayesian network. With the Standard Assessment Tool, you can specify full and causally independent probability distributions. With the Asymmetric Assessment Tool, you can avoid specifying redundant probabilities. If you have sufficient data and use machine learning tools to create Bayesian networks, you can use MSBNx to edit and evaluate the results. MSBNx is fully component based. Its most important component is MSBN3, an ActiveX DLL. MSBN3 offers an extensive COM-based API for editing and evaluating Bayesian networks. You'll find MSBN3 especially easy to use from COM-friendly languages such as Visual Basic and JScript. A comprehensive manual, available at  HYPERLINK "http://research.microsoft.com/adapt/MSBNx/" http://research.microsoft.com/adapt/MSBNx/ documents MSBN3. MSBNx also includes graphical components, for example, both the Standard Assessment and Asymmetric Assessment tools are ActiveX controls and can be used in other applications. Also, you can extend the editing and evaluation abilities of MSBNx by creating add-ins. For example, MSBNx ships with an add-in for editing and evaluating Hidden Markov Models. This report starts with an introduction to tasks of diagnosis and troubleshooting and a review of the concepts and vocabulary of Bayesian Networks. It then outlines the features of MSBNx. Next, it describes key aspects of the toolkit, including the inference and modeling environment. Finally, the report illustrates with examples how other programs can use MSBNxs inference and modeling services. Background on Diagnosis and Troubleshooting This section provides background information about on the tasks of diagnosis and troubleshooting with Bayesian networks [Breese & Heckerman, 1996; Heckerman, Horvitz, & Nathwani, 1992; Heckerman & Breese, 1994a; Heckerman & Breese, 1994b; Horvitz, Breese, & Henrion, 1988]. Consider this example of troubleshooting from real life. When your car doesn't start, what should you do? A) Check if your lights work B) Replace your fuel pump and see if that helps Choice A) seems wiser. Why? Because checking your lights tells you much about your car's battery, a likely cause of your problem. But that's not the whole reason to check your lights first. Just as important, you check your lights first because such a check is easy and inexpensive. MSBNx is a tool for doing this kind of cost-benefit reasoning for diagnosis and troubleshooting. Causes and Uncertainty: You start using MSBNx by creating a model of your system. First, you create a diagram showing what causes what. For example, in our car model, the battery power causes the lights to on. Battery power also causes the engine to turn over which, in turn, causes the car to start. Second, you tell about the uncertainty in the system, giving your probabilities for various situations. Here are some examples: Before you tried to start your car, how likely would you find a bad fuel pump? In our car model we set this to 2%. If battery power is high, how likely are the lights to work? We set this to 99.9% Third and finally, you can add cost information. In the car model, we set the cost of checking the lights at $1. We say that fuel pumps can't be tested, only replaced, which costs $100. Your model of causes and effects is called Bayesian Network (or sometimes Bayes' Net or Belief Network). Run It Backwards: Once you've created a model, you can use it for diagnosis and troubleshooting. MSBNx evaluates your causes-to-effects model by running it backwards from effects to causes. Given known effects, for example, evidence of the lights working, it computes the probability of causes such as a bad battery. It also computes the utility of further tests, for example, checking the gas gauge. Although, MSBNx recommends which test to do next, you can do tests in any order. As you enter new evidence, MSBNx efficiently recomputes the probability of possible causes and the recommended order of further tests. Basics of Bayesian Inference and Bayesian Networks This section provides background information about Bayesian Networks [Jensen, 1996; Pearl, 1988, Poole, 1998]. Motivation Logic, both in mathematics and in common speech, relies on clear notions of truth and falsity. Information that is either true or false is known as Boolean logic. For example, consider a statement such as "Unless I turn the lights on, the room will be dark." It leaves no room for uncertainty. However, what about daylight coming in the windows? Everyday life presents us with many situations in which the accumulation of evidence leads to a conclusion. For example, a small amount of oil underneath a car in your driveway may be no great cause for concern. You may remember that you recently changed the car's oil, spilling some on the driveway in the process. If so, the oil stain evidence may be unimportant; if not, you may have your car checked for an oil leak, or wait to see if another oil stain shows up underneath your car when you park it in a different area. Bayesian probability theory is a branch of mathematical probability theory that allows one to model uncertainty about the world and outcomes of interest by combining common-sense knowledge and observational evidence. Bayesian networks A Bayesian network is: a set of variables, a graphical structure connecting the variables, and a set of conditional distributions. A Bayesian network is commonly represented as a graph, which is a set of vertices and edges. The vertices, or nodes, represent the variables and the edges, or arcs, represent the conditional dependencies in the model. The absence of an arc between two variables indicates conditional independence; that is, there are no situations in which the probabilities of one of the variables depends directly upon the state of the other. Construction of a Bayesian network follows a common set of guidelines: Include all variables that are important in modeling your system. Use causal knowledge to guide the connections made in the graph Use your prior knowledge to specify the conditional distributions. Causal knowledge in this context means linking variables in the model in such a way that arcs lead from causes to effects. What is a Discrete Variable? A variable is an element of a probability model that can take on a set of different values that are mutually exclusive and exhaustive. Consider a medical experiment in which men and women of different ages are studied. Some of the relevant variables would be the sex of the participant, the age of the participant and the experimental result. The variable sex has only two possible values: male or female. The variable age, on the other hand, can take on many values. While probability theory as a whole can handle variables of both types, MSBNx can only accept variables with a limited number of possible values. We call these discrete variables, and we call each of the set of possible values a state. How would you handle the age variable in MSBNx? Most likely, you would create a variable where each state represents a range of years. For example: Child (0-18) Adult (18-45) Middle Aged (45-65) Senior (65+) This process is known as discretization. Relationships in a Bayesian Model In probability theory, there is no a priori way of knowing which variables influence other variables. In general, the complete or joint probability distribution must be known to correctly perform inference. For a real-world model, the joint distribution is usually very large and cannot be directly stored on a computer. One of the primary roles of a Bayesian model is to allow the model creator to use commonsense and real-world knowledge to eliminate needless complexity in the model. For example, a model builder would be likely to know that the time of day would not normally directly influence a car's oil leak. Any such influence would be based on other, more direct factors, such as temperature and driving conditions. The method used to remove meaningless relationships in a Bayesian model is to explicitly declare the meaningful ones. After establishing all the variables in a model, you must deliberately associate variables that cause changes in the system to those variables that they influence. Only those specified influences are considered. These influences are represented by conditioning arcs between nodes. Each arc should represent a causal relationship between a temporal antecedent (known as the parent) and is later outcome (known as the child). What is Inference? Inference, or model evaluation, is the process of updating probabilities of outcomes based upon the relationships in the model and the evidence known about the situation at hand. When a Bayesian model is actually used, the end user applies evidence about recent events or observations. This information is applied to the model by "instantiating" or "clamping" a variable to a state that is consistent with the observation. Then the mathematical mechanics are performed to update the probabilities of all the other variables that are connected to the variable representing the new evidence. After inference, the updated probabilities reflect the new levels of belief in (or probabilities of) all possible outcomes coded in the model. These beliefs are mediated by the original assessment of belief performed by the author of the model. The beliefs originally encoded in the model are known as prior probabilities, because they are entered before any evidence is known about the situation. The beliefs computed after evidence is entered are known as posterior probabilities, because they reflect the levels of belief computed in light of the new evidence. Parents and Children: Introduction to Diagrams To recap, every variable in the real world situation is represented by a Bayesian variable. Each such variable describes a set of states that represent all possible distinct situations for the variable. Once the set of variables and their states are known, the next step is to define the causal relationships among them. For any variable, this means asking the questions: What other variables (if any) directly influence this variable? What other variables (if any) are directly influenced by this variable? In a standard Bayesian network, each variable is represented by a colored ellipse; this graphical representation is called a node. Each causal influence relationship is described by a line (or arc) connecting the influencing variable to the influenced variable. The influence arc has a terminating arrowhead pointing to the influenced variable. The common terminology then is as follows: A node is a Bayesian variable. An arc connects a parent (influencing) node to a child (influenced) node. Bayesian network Creation To create a Bayesian network, then, the following steps are necessary. Create a set of variables representing the distinct elements of the situation being modeled. For each such variable, define the set of outcomes or states that each can have. This set is referred to in the mathematical literature as "mutually exclusive and exhaustive," meaning that it must cover all possibilities for the variable, and that no important distinctions are shared between states. Establish the causal dependency relationships between the variables. This involves creating arcs (lines with arrowheads) leading from the parent variable to the child variable. Assess the prior probabilities. This means supplying the model with numeric probabilities for each variable in light of the number of parents the variable was given in Step 3. Running Inference After the model has been developed, MSBNx allows you to model or mimic real-world situations by entering or removing evidence on the model. As you do so, there are various methods for displaying the resulting probabilities. Recommendations and Troubleshooting When inference is performed on a model, there are various mathematical schemes for discovering which pieces of evidence would be the most important to discover. The algorithms or procedures produce a list of variables that have not been given evidence. Simply put, they are ordered by the degree to which the application of evidence would simplify the situation. MSBNx Features This section outlines the most interesting features of MSBNx. The details of the features can be found in later sections of this report and on the MSBNx web pages ( HYPERLINK "http://research.microsoft.com/adapt/MSBNx/" http://research.microsoft.com/adapt/MSBNx/). MSBNx offers these features: Graphical Editing of Bayesian networks Use MSBNx to graphically relate causes to effects. Express a system's uncertainty by giving your probabilities of various situations. Exact Probability Calculations MSBNx uses clique-tree propagation methods to calculate the probabilities exactly. Decision-Theoretic Diagnosis, Troubleshooting, and Recommendations Using your cost information or defaults, MSBNx dynamically recommends troubleshooting steps. It bases its recommendations on a cost-benefit analysis. If you provide no cost information, a Value of Information (VOI) measure determines the recommendation order. XML Format MSBNx loads and stores files in an XML-based format. COM API The MSBN3 ActiveX DLL provides an COM-based API for editing and evaluating Bayesian networks. MSBN3s event-based design makes it especially easy to use from COM-friendly languages such as Visual Basic and JScript. Comprehensive Help Pages Help pages for both the MSBNx editor and the MSBN3 API are on-line The manuals are also installed as Windows Help documents. While editing with MSBNx or while programming with MSBN3, pressing F1 will bring up help for the topic on which you are currently working. The MSBN3 API Help includes overviews of loading models, running inference, and other common programming tasks. Other ActiveX controls MSBNx includes several graphic components that can be used directly by other programs written in languages that support ActiveX, such as Visual Basic, Jscript, and C++. Add-Ins You can extend the editing and evaluation abilities of MSBNx by creating add-ins. These add-ins are ActiveX DLLs. They can be created in languages such as Visual Basic. Standard Assessment of Probabilities Use the Standard Assessment Tool to assess the probabilities of various situations. Reduce the number of probabilities you need to specified by assuming causal independence. Control the formatting of the probability tables. Optionally display probabilities as odds. Use the Standard Assessment Tool as an ActiveX component in other programs. Asymmetric Assessment When different situations can be treated the same, you can use the Asymmetric Assessment Tool to reduce the number of probabilities you need to specified. Dynamic Properties Either interactively or programmatically, you can define and attach property values to models and nodes. For example, you can define a property type called DatabaseKey with string values. Then on each node in a model, you can attach a string value to used for looking up more information related the node in a database. Model and nodes can have any number of properties. Property values can be strings, reals, arrays of string or reals, or enumerated types. Work with Machine-Learned Models The WinMine Toolkit ( HYPERLINK "http://research.microsoft.com/~dmax/WinMine/Tooldoc.htm" http://research.microsoft.com/~dmax/WinMine/Tooldoc.htm) uses machine learning and statistical methods to create Bayesian networks. MSBNx can load, edit, and evaluate the models created by WinMine. Hidden Markov Models MSBNx includes an (undocumented) add-in for editing and evaluating Hidden Markov Models (HMMs). Inference on HMMs is exact, but inefficient for large models. The MSBNx Application MSBNx is a Microsoft Windows software application. In MSBNx each model is represented as a graph or diagram. The random variables are shown as ellipses, called nodes, and the conditional dependencies are shown as arrows, or directed arcs, between variables. At the present time, MSBNx only supports discrete distributions for its model variables. MSBNx supports simultaneous viewing and evaluation of multiple models. The main tasks within MSBNx are  HYPERLINK "http://msr-webdev/adapt/msbnx/Model_Creation.htm" Model Creation -- creating and modifying models and their diagrams. Working with Model Diagrams  HYPERLINK "http://msr-webdev/adapt/msbnx/Model_Evaluation.htm" Model Evaluation -- control the probabilistic inference process.  HYPERLINK "http://msr-webdev/adapt/msbnx/Assessment_Overview.htm" Probability Assessment -- enter and maintain prior probabilities. Lets look at each of these in turn. Model Creation Models are created in MSBNx as new model documents. Once the empty document window appears, you can add variables and dependency arcs for your model. By default, variables (or nodes) are created without prior probability distributions. You can easily and automatically supply uniform distributions of different kinds at any time. After the applications loads, you can click the "new document" button on the main application toolbar. An empty diagram document is created and given the default name. To create the elliptical nodes that represent random variables, right click anywhere in the diagram to activate the diagram context menu. Choose "New Node..." from the menu. An edit box will appear on the diagram with a default name for the new variable and you can enter a new name. When you strike the Enter key, the new node appears. To create a conditioning arc between two nodes, do the following steps. Click the left mouse button anywhere on the canvas. This de-selects any selected nodes or other objects. Position the mouse button in the ellipse representing the parent variable. While holding down the control (CTRL) key, also hold down the left mouse button and drag the mouse into the node representing the child variable. By default, MSBNx will create a discrete distribution for every node created. You can change this behavior using the diagram window context menu. You have the choice of automatically creating standard discrete distributions, causally independent distributions or no distribution. When dependency arcs are added or removed, the dimensionality of the distribution of the child node is changed accordingly. Working with Model Diagrams The model diagram window is the primary document of MSBNx. Each window represents a Bayesian network. Each window can only view a single Bayesian network, but multiple windows can view the same Bayesian network. You can control the diagram window by Direct manipulation of the objects within it, using standard operations such as dragging and selection. Choosing enabled buttons from the right-hand toolbar. Choosing menu items from context menus available on the diagram. When maximized, a typical diagram window might appear as:  INCLUDEPICTURE "http://msr-webdev/adapt/msbnx/mainwind3.jpg" \* MERGEFORMATINET  As with most modern graphical applications, you can simply left-click you mouse on an object. Much of the functionality of MSBNx and its components is presented through context menus. To see a context menu for the diagram, right-click the mouse over any unoccupied space in the diagram. To see a context menu for a node (variable) or arc (edge), right-click on it. Inference can be performed directly from a diagram window. To enter evidence for a variable, use its node's context menu to set its state. To view the posterior probabilities, enable bar chart displays for any or all of the nodes in the model. Model Evaluation The Model Evaluation window is the primary means of performing inference using a model. (An alternate method is to use bar charts on the main diagram window.)  INCLUDEPICTURE "http://msr-webdev/adapt/msbnx/EvalWind.jpg" \* MERGEFORMATINET  Performing inference means manipulating evidence in a model and obtaining the posterior probabilities resulting from the changes in evidence. There are two primary sections to the Evaluation window. The left-hand pane contains a tree-based view of the nodes in the model. The right-hand pane contains a tabbed display of the results of inference presented in several ways. The basic Evaluation window has the results for a single model. The form can be extended to contain several pairs of evaluation panes (tree view and tabs), each of which can have a different appearance or represent a different model. The primary purposes of the tree view are to allow you to choose which model variables are shown in the results pane and to enter or remove evidence from the model's nodes. Clicking in the black rectangle next to a node's name causes the posterior probabilities of that node to appear in the spreadsheet and bar chart views to the right of the tree. Right-click on the name of a node and a menu appears containing state names for the node, allowing you to change the evidence. By default, the Tree View has a single root level, named after the model itself. In the default view, all the nodes of the model are attached directly to this level. You can also use node properties to create another level of organization in the tree view. See Evaluation Display Options for more information. The right-hand section of an Evaluation window normally has a set of tabs across its top. (This can be controlled using the Evaluation Display Options.) When the Spreadsheet tab is active, posterior probabilities are shown for the nodes that are checked in the tree view. Each pair of rows represents a node, and each column represents a state of that node. When the Bar Chart tab is active, a vertical colored bar appears for each node checked in the tree view. The probability mass associated with each state is represented as a distinct color band in the bar. During evaluation, models configured to support troubleshooting or diagnostic recommendations can provide an ordered list of most effective diagnostic steps. The Recommendations tab shows an ordered list of nodes (by name and description). Each set of lines in the list shows the information score (cost in dollars) of the node and each state of the node. The nodes are shown in order of least cost. Probability Assessment MSBNx supports these kinds of probability distributions: Discrete Sparse. This is the standard type of distribution. All possible probabilities are available, but there is no requirement that all values be specified. During evaluation, uniform distributions are automatically supplied for unassessed probabilities. Causally Independent. This type of distribution compresses the space of necessary probabilities by assuming that certain states of the parent nodes are mutually exclusive. CI distributions can be changed to Discrete Sparse distributions without loss. MSBNx supports these kinds of probability assessment: Standard assessment is a table-based method. You can locate and edit a particular set of probabilities based upon the states of the parents of the variable. Causally Independent or CI assessment is a type of standard discrete assessment. It uses a special type of distribution based upon assumptions about conditional independence among the parents of a variable. Such distributions can be very convenient since they dramatically reduce the number of values that must be entered. Asymmetric assessment is a tree-based method. You create sets of probabilities, organized as a decision tree, by making explicit distinctions between states of parents of the variable. The tree serves to reduce the number of values that must be entered, since entire branches can easily be given common values. You cannot perform asymmetric assessment on variables with causally independent distributions. The mechanics of each of these assessment types have one goal in common: to minimize the number of distinct probabilities required to correctly specify the distribution. Creating Models for Troubleshooting and Diagnosis A common use of Bayesian network models is to diagnosis system failures in a probabilistic framework. This has several advantages over conventional rule-based or decision-tree methods, since Bayes nets support uncertain evidence in a theoretically correct fashion. In addition, prior distributions in Bayes nets can be built that model logical functions such as AND, OR and NOT using what are known as deterministic nodes; that is, nodes whose distributions contain only zeroes and ones. Such nodes, therefore, act as logic gates. A key question in decision theory is this: In the current evidence setting, what new evidence would most effectively lead to a clear diagnosis? Often known as the value of information, information theory provides mathematical approaches to answering this question. Types of Decision-Theoretic Diagnosis MSBNx supports two algorithms that use information theory to order or rank variables in a Bayes net according to their information weight or influence. In either scenario, variables or nodes in the model play certain roles. These roles are also known as labels, and must be assigned correctly or the results cannot be interpreted. Both methods produce as a result an ordered list of variables ranked by a value of information score. In a typically implementation, for example, this list would determine the order of questions being asked of a diagnostician or technician. Diagnosis: Mutual Information In such a model, variables are assigned one of two roles: Hypothesis Node. Also known as a hidden variable, this is typically a variable that cannot be directly observed. It is the target or purpose of the overall diagnosis. Information Node. An observable variable that influences the hypothesis node(s) in the model. There may be other nodes in the model which are not labeled; although they influence inference in the normal way, they do not otherwise enter into the diagnostic process. Utility-based diagnosis uses mutual information to compute the amount of weight or "lift" that evidence about the state of each information node would bring to each hypothesis variable. The resulting ranking of uncertain (undetermined) information nodes is used to expedite the diagnostic process. Troubleshooting: Fix-or-Repair Planning In addition to being assigned to roles, variables in a troubleshooting model are also given one or more costs. The Bayesian network author may consider that these costs are measured in dollars (or other monetary currency), time (in minutes or seconds) or any other unit that is consistent with the problem formulation. Troubleshooting uses an algorithm that iterates over all reasonable repair plans in an attempt to find the ones with the highest likelihood of success at the cheapest cost. The result is a list of nodes, ordered by cost. Establishing evidence about the top-ranked (cheapest) node is guaranteed (within the limits of the model) to lead to correct diagnosis in the shortest and cheapest number of steps. Requirements for Diagnosis To perform mutual information diagnosis in a model: At least one node in a diagnostic model must be identified as an hypothesis node At least two of its nodes must be identified as information nodes. Requirements for Troubleshooting Cost Factors There are three types of costs that are important in a troubleshooting model. Cost to Observe. This is the cost of observing a symptom or sensor. For example, testing the battery of a car or running a blood test for gram-positive bacteria. Cost to Fix. This is the cost of fixing or replacing a component in a system. For example, replacing the power supply in a computer. Service cost. This is a cost assigned to the network or model as a whole. In other words, it indicates the cost that would be expected if the diagnostic operation failed. For example, if a computer server in a network could not be repaired through diagnosis it would have to be replaced. Each of the different possible roles of variables in troubleshooting networks may have either a cost to observe, a cost to fix, both or neither. The service cost of the model is treated as the cost to fix for the entire model as a whole. Note: The service cost is required to perform troubleshooting. The roles of variables in a troubleshooting model and their costs are as follows NameCosts AllowedPurposeinformationalobserveUsed to define observable evidentiary variablesproblem-definingfixUsed to define primary symptoms of failure; that is, the element of the model that is the target of the diagnosis.fixable and observableobserve and fixUsed to define observable and replaceable elementsfixable but unobservablefixUsed to define elements that can only be replaced or repairedunfixableneitherUsed to define elements that can neither be fixed or observedotherneitherUsed to define variables that play no direct part in the diagnostic process. These may be deterministic or "modal" variables that reshape the problem in a logical fashion.Establishing "Problem" Nodes The most vital part of a troubleshooting network is its problem nodes. These nodes must be declared in a particular manner: state zero (the first state declared) must be associated with the normal behavior of the component or element. All other states must be associated with the mutually exclusive and exhaustive set of states associated with failure modes of the component. Many problem nodes have only two states: "Works" and "Doesn't Work". If a problem node has more than two states, they must correlate to clearly distinct situations. Consider a computer printer with four states: "Works", "No Paper is Output" "Printing is Very Slow", and "Printing is Garbled". In each case, observation allows its problem state to be distinguished. (There is, however, some ambiguity-- consider a case where printing is both garbled and slow.) Multiple problem nodes may be defined, but only one is actually considered during any given troubleshooting session. Using Troubleshooting The mechanics of troubleshooting work as follows. One of the problem nodes is set (instantiated) to one of its problem states (that is, a state other than state zero). The Troubleshooting Recommendations algorithm is run, and a ranked list of nodes is returned, each with its predicted utility. The highest (first) variable in the ranked list is the one with the lowest cost. The technician or diagnostician would then attempt to gather evidence about this variable. The evidence found about the highest ranked variable is entered as evidence into the model. Alternatively, evidence can be entered for any other uninstantiated node in the collection. Return to step 2. Evaluation of Diagnostic and Troubleshooting Models The evaluation window of MSBNx will attempt to determine the correct type of procedure to perform. The rules it uses are as follows. If there is a node labeled as problem-defining, then troubleshooting diagnosis is enabled. If there is a node labeled as hypothesis, then mutual information diagnosis is enabled. Otherwise, diagnosis is disabled. If other criteria for diagnosis are not met, an error message will indicate the situation. For additional information see the references. The MSBN3 Component MSBN3 is an ActiveX DLL that offers editing and inference services to other programs such as MSBNx. Because MSBN3 fires events whenever a model is edited, programs that use MSBN3 can be event-driven. This allows the programs to be more modular; indeed, multiple stand-alone programs, add-ins, and utilities to work on the same model without having to know about each other. MSBN3 Object Model The figure shows the MSBN3 object model (except those objects related directly to  HYPERLINK "http://msr-webdev/adapt/msbn3/Overviews/InferenceModel.htm" \o "Inference Model" inference).  INCLUDEPICTURE "http://msr-webdev/adapt/msbn3https://www.microsoft.com/images/MainObjectModel1.gif" \* MERGEFORMATINET  For example, the figure shows that: the  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/msbn.htm" MSBN object contains  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Models.htm" Models collection, the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Models.htm" Models collection is made up of  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Model.htm" Model objects, a  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Model.htm" Model object contains a  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/ModelNodes.htm" ModelNodes collection, a  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/ModelNodes.htm" ModelNodes collection is made up of  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Node.htm" Node objects, a  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Node.htm" Node object contains a  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Properties.htm" Properties collection, a  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Properties.htm" Properties collection is a mapping from  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/PropertyType.htm" PropertyType objects to variant values. Loading and Saving Models This first task for any application that uses MSBN3 is to load or create a model. Other common operations are copying, renaming, removing, and saving. This section illustrates the operations by annotating a simple Visual Basic program. Creating a MSBN object and Loading a Model Here we create a new MSBN object. Then we use the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Models.htm" Models collection's  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htm" Add method to load file "auto.dsc" and call it "Auto". Any error message from the loading go into file "errorfile.log". ' Creating a new MSBN object Dim aMSBN As New MSBN3Lib.MSBN 'Loading a model Dim model1 As MSBN3Lib.Model Set model1 = aMSBN.Models.Add("Auto", FileName:="auto.dsc", ErrorFilename:="errorfile.log") Creating an Empty Model An empty model also created with  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htm" Add, but without the FileName argument. 'Creating an empty model Dim model2 As MSBN3Lib.Model Set model2 = aMSBN.Models.Add("Model 2") Renaming and Loading a Model To rename a model, change the value of its  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Name.htm" Name property. 'Rename it model2.Name = "cat" The  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Load.htm" Load method on a model will replace the current contents of the model with the contents of a file. If a FileName is given, that file will be used. If no FileName is given, the last filename associated with the file (given by the Model's  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/FileName.htm" FileName property) is used. The  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Load.htm" Load method can include the name of an error file for messages generated during the load. 'Load model2.Load FileName:="cat.dsc", ErrorFilename:="errorfile.log" Saving a Model The  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Save.htm" Save method on a model will save the contents of the model to a file. A FileName and FileFormat can specified. If either is not specified then their last values (found in the properties  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/FileName.htm" FileName and  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/FileFormat.htm" FileFormat) are used. 'Save the auto it to a different place and format model1.Save FileName:="c:\temp\Auto0.xbn", FileFormat:=fileformat_Xml Loading the Same Model Twice The  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Models.htm" Models collection can not have two models with the same name, so here we don't specify a name. This causes a new unique name to be generated and allows us to load the auto model twice. 'Load in another copy of the auto model aMSBN.Models.Add FileName:="auto.dsc", ErrorFilename:="errorfile.log" Copying a Model The  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Copy-Model.htm" Copy method applied to a model, returns a copy of that model. The copy is also added to the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Models.htm" Models collection. If no name is given, the copy will have a generated name. 'Copy the cat model model2.Copy Listing All Models Like any collection, the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Models.htm" Models collection can be enumerated (see  HYPERLINK "http://msr-webdev/adapt/msbn3/Overviews/Collections.htm" Collections and Maps: an Overview in the MSBN3 manual). Its  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Description.htm" Description property can create a string listing the name of all its models. For example: 'Print the name of the all models Debug.Print aMSBN.Models.Description produces "Auto,cat,Model(1),cat(2)". Removing Models As with any collection, elements of the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Models.htm" Models collection can be removed with the  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Remove.htm" Remove method. See  HYPERLINK "http://msr-webdev/adapt/msbn3/Overviews/Collections.htm" Collections and Maps: an Overview in the MSBN3 manual for more information. 'Remove the cat model aMSBN.Models.Remove "cat" Inference MSBN3's primary task is inference, in other words, when told some things about the world it estimates the probability of other things in the world. It can also recommend what additional information would be most valuable. This object model shows the objects related to inference.  INCLUDEPICTURE "http://msr-webdev/adapt/msbn3https://www.microsoft.com/images/InferenceModel1.gif" \* MERGEFORMATINET  For example it shows that: a  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Model.htm" Model object contains an  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Engine.htm" Engine object, an  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Engine.htm" Engine object contains an  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Evidence.htm" Evidence collection, an  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Evidence.htm" Evidence collection is a mapping from  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Node.htm" Node objects to  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/State.htm" State This section illustrates inference by annotating a simple Visual Basic program. Loading a Model and Accessing Its Inference Engine Here we load a model for troubleshooting car problems. A model's inference engine is available from its  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Engine.htm" Engine property. Dim aMSBN As New MSBN3Lib.MSBN 'Load the Auto model Dim modelAuto As MSBN3Lib.Model Set modelAuto = aMSBN.Models.Add("Auto", FileName:="auto.dsc", ErrorFilename:="loaderror.log") 'Call the model's inference engine "inferAuto" Dim inferAuto As MSBN3Lib.Engine Set inferAuto = modelAuto.Engine Finding a Probability This model has a node named "EngineStart" that has a state named "yes". To find the probability that the car's engine will start, we apply the inference engine's  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Belief.htm" Belief function to the node and state of interest. ' Check a probability Debug.Print "The probability that the car will start is " Debug.Print inferAuto.Belief("EngineStart", "yes") Debug.Print Output: The probability that the car will start is 0.899733245651341 To assert a fact, we add a node and state to the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Evidence.htm" Evidence collection using the  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Add-Evidence.htm" Add method: ' Add some evidence and check some probabilities Debug.Print "Tell it that the car, in fact, doesn't start." inferAuto.Evidence.Add "EngineStart", "no" Now, if we ask for the probability that the car will start it will say 0. We can also ask about other probabilities. Debug.Print "The probability that the car will start is " Debug.Print inferAuto.Belief("EngineStart", "yes") Debug.Print "The probability that the battery is good is " Debug.Print inferAuto.Belief("Battery", "good") Debug.Print Output: Tell it that the car, in fact, doesn't start. The probability that the car will start is 0 The probability that the battery is good is 0.990498642853832 We'll tell it that the lights don't work and ask about the probability that the battery is good. ' Add some more evidence and check some probabilities Debug.Print "Tell it the lights don't work" inferAuto.Evidence.Add "LightsShine", "don't work" Debug.Print "The probability that the battery is good is " Debug.Print inferAuto.Belief("Battery", "good") Debug.Print Output: Tell it the lights don't work The probability that the battery is good is 0.97598617912302 Using the  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Set-Assignment-Evidence.htm" Set method on the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Evidence.htm" Evidence collection, we can change the evidence and say that lights do work. ' Change some evidence and check some probabilities Debug.Print "Tell it the lights do work" inferAuto.Evidence.Set "LightsShine", "work" Debug.Print "The probability that the battery is good is " Debug.Print inferAuto.Belief("Battery", "good") Debug.Print Output: Tell it the lights do work The probability that the battery is good is 1 By enumerating every node and every state, we can see every probability: 'For every node and every state, show the probability Dim aNode As MSBN3Lib.Node Dim aState As MSBN3Lib.state Debug.Print "Nodes and probabilities" For Each aNode In modelAuto.ModelNodes Debug.Print aNode.Name, For Each aState In aNode.States Debug.Print "P("; aState.Name; ")="; inferAuto.Belief(aNode, aState), Next aState Debug.Print Next aNode Debug.Print Output: Nodes and probabilities Alternator P(good)= 0.999989576605324 P(bad)= 1.0423394676112E-05 Battery P(good)= 1 P(bad)= 0 BatteryPower P(good)= 1 P(low)= 0 P(none)= 0 CD P(yes)= 0.999957618165044 P(no)= 4.23818349563663E-05 CDLeak P(good)= 0.999998957660463 P(bad)= 1.0423395366561E-06 Distributor P(good)= 0.841989863011888 P(bad)= 0.158010136988112 ESLeak P(good)= 0.984198985254525 P(bad)= 1.58010147454754E-02 ETO P(yes)= 0.83456582766937 P(no)= 0.16543417233063 ETOLeak P(good)= 0.984946844194984 P(bad)= 1.50531558050155E-02 EngineStart P(yes)= 0 P(no)= 1 FanBelt P(ok)= 0.999968729816554 P(loose)= 3.12701834458835E-05 FuelLine P(good)= 0.841989863011888 P(bad)= 0.158010136988112 FuelPump P(good)= 0.699067412143921 P(bad)= 0.300932587856079 Gas P(not empty)= 0.991603194714833 P(empty)= 8.39680528516724E-03 GasGauge P(not empty)= 0.990611591485833 P(empty)= 9.3884085141671E-03 LightsShine P(work)= 1 P(don't work)= 0 RadioPlays P(works)= 0.998999999965425 P(doesn't work)= 1.00000003457535E-03 SparkPlugs P(good)= 0.774046752305156 P(bad)= 0.225953247694844 Starter P(good)= 0.849468451921102 P(bad)= 0.150531548078898 Recommendations Use the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm" Recommendations collection to find how useful additional information about a node would be. The first item in  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm" Recommendations. HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Keys.htm" Keys is the name of the most valuable node. 'Show a recommendation Debug.Print "Knowing the state of "; inferAuto.Recommendations.Keys(0); "would be most useful." Debug.Print Output: Knowing the state of ETOwould be most useful. Enumerating the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm" Recommendations collection (and its  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm" KeyObjects collection) list all the useful nodes and their utility (in order of utility). 'List all recommendations Debug.Print "Node", "Utility" For Each aNode In inferAuto.Recommendations.KeyObjects Debug.Print aNode.Name, inferAuto.Recommendations(aNode) Next aNode Output: Node Utility ETO -166.471758124647 GasGauge -203.17297015686 SparkPlugs -205.971769894314 Distributor -206.95861573907 Gas -216.904853682089 FanBelt -220.954647059405 Alternator -220.965951259178 FuelPump -234.409330429375 ESLeak -247.667290007677 ETOLeak -248.025747382471 FuelLine -252.400377437429 CDLeak -255.971139079536 Battery -285.971769894314 Starter -335.685042698628 Distributions Before MSBN3 can do inference, it must know the conditional probability for every node given its parents. This probability distribution is specified has the  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist property of every node. If a node does not have a distribution, it's  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist property will be Nothing and no inference will be possible. This overview tells how these conditional distributions are created, read, and edited. Distributions as Tables If all nodes are discrete (as in the case currently in MSBN3), a conditional distribution between a node and its parents can be thought of as a table. For example, suppose we have this model of a cat:  INCLUDEPICTURE "http://msr-webdev/adapt/msbn3/Overviews/Cat.gif" \* MERGEFORMATINET  in which a cat's happiness is caused by being feed, being petted, and seeing a bird. The conditional distribution of Happy given Feed, Petted, and SawBird can be thought of as a table. Suppose it has these values: Feed Petted SawBirdHappyYesNoYesYesYes0.9719090.0280912YesYesNo0.8757260.124275YesNoYes0.8777860.122214YesNoNo0.598550.40145NoYesYes0.50.5NoYesNo0.08877890.911221NoNoYes0.2486260.751374NoNoNo0.01877860.981221The table has one row for each possible assignment of parent nodes. It also has one column of parameter values for each state in the Happy node. Accessing Parameter Values The parameter values of a distribution can be accessed via the  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist object by specifying a row and column. For example, if nodeHappy is the node in the model with name "Happy", then this expression: nodeHappy.Dist(0, "Yes") will return value 0.971909. Parameters can also be set this way. For example, this expression changes the parameters of the last row: nodeHappy.Dist(7, "Yes") = 0.01 nodeHappy.Dist(7, "No") = 0.99 As the examples show, rows can be specified with an integer index. Additionally, they can be specified with a parent assignment. This code creates a parent assignment and uses it to change the parameters. Dim anAssign As MSBN3Lib.Assignment Set anAssign = modelCat.CreateAssignment anAssign.Add "Feed", "No" anAssign.Add "Petted", "Yes" anAssign.Add "SawBird", "Yes" ' anAssign is now: Feed->No,Petted->Yes,SawBird->Yes nodeHappy.Dist(anAssign, "Yes") = 0.55 nodeHappy.Dist(anAssign, "No") = 0.45 The parent assignment must include a setting for every parent, but the parents can be in any order and other nodes can be included, too. (Other nodes will just be ignored.) Table Dimensions The  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist object's  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Count.htm" Count property tells the number of rows in the table. nodeHappy.Dist.Count The number of columns is available from these equivalent expressions: nodeHappy.States.Count nodeHappy.Dist.Node.States.Count Parent Assignments for each Row To find the parent assignment of a given row, use the  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist object's  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm" KeyObjects property. It takes a index integer and returns the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Assignment.htm" Assignment collection for the row. (Like other MSBN3 collections,  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist. HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm" KeyObjects also accepts objects, but unlike other collections, it does not accept strings.) This expression retrieves the parent assignment for the row with index 2 (the third row) and then turns that  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Assignment.htm" Assignment collection into a string nodeHappy.Dist.KeyObjects(2).Description The returned value is the string "Feed->Yes,Petted->No,SawBird->Yes". Similarly, this expression retrieves the same parent assignment and then finds the state to which the Petted node is set: nodeHappy.Dist.KeyObjects(2)("Petted").Name It returns "No". The documentation for the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Assignment.htm" Assignment collection has details on access and enumeration of  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Assignment.htm" Assignment collections. Enumerating the Rows The rows of a distribution can be enumerated with  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist object's  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm" KeyObjects collection. For example, this code fragment prints parent assignments and parameter values. For Each anAssign In nodeHappy.Dist.KeyObjects Debug.Print anAssign.Description, For Each aState In nodeHappy.States Debug.Print nodeHappy.Dist(anAssign, aState), Next aState Debug.Print Next anAssign Output: Feed->Yes,Petted->Yes,SawBird->Yes 0.971908986568451 0.028091199696064 Feed->Yes,Petted->Yes,SawBird->No 0.875725984573364 0.124274998903275 Feed->Yes,Petted->No,SawBird->Yes 0.877785980701447 0.122213996946812 Feed->Yes,Petted->No,SawBird->No 0.598550021648407 0.401450008153915 Feed->No,Petted->Yes,SawBird->Yes 0.55 0.45 Feed->No,Petted->Yes,SawBird->No 8.87788981199265E-02 0.911221027374268 Feed->No,Petted->No,SawBird->Yes 0.248625993728638 0.751374006271362 Feed->No,Petted->No,SawBird->No 0.01 0.99 Default Parameters Suppose we want to model the happiness of a second cat with this distribution: Feed Petted SawBirdHappyCat2YesNoYesYesYes0.990.01YesYesNo0.010.99YesNoYes0.010.99YesNoNo0.010.99NoYesYes0.010.99NoYesNo0.010.99NoNoYes0.010.99NoNoNo0.010.99This cat is only likely be happy when everything is going its way. Do we really have to specify identical parameters for 7 of the 8 rows? The answer is no. We can instead, specify default parameters and let 7 of the 8 rows use that default. This code fragment shows the creation of a new node with parents and states based on nodeHappy. The  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/AddDist.htm" AddDist method is used to add a distribution to the node. The default parameter values of the distribution are then set with  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Default.htm" Default. These at first apply to all rows, but then are over ridden for row #0. We use  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/UsingDefault.htm" UsingDefault to confirm that row #0 is not using the default parameters but that row #1 is. ' Create a node for a new cat's happiness. Dim nodeHappyCat2 As MSBN3Lib.Node Set nodeHappyCat2 = modelCat.ModelNodes.Add("HappyCat2", "HappyCat2", _ States:=nodeHappy.States, ParentNodes:=nodeHappy.ParentNodes) nodeHappyCat2.AddDist ' Specify the default parameters nodeHappyCat2.Dist.Default("Yes") = 0.01 nodeHappyCat2.Dist.Default("No") = 0.99 ' Override the default on row #0 nodeHappyCat2.Dist(0, "Yes") = 0.99 nodeHappyCat2.Dist(0, "No") = 0.01 'Confirm that row #0 is not using the default and that row #1 is: Debug.Assert Not nodeHappyCat2.Dist.UsingDefault(0) Debug.Assert nodeHappyCat2.Dist.UsingDefault(1) The  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/UsingDefault.htm" UsingDefault property can also be set true or false to force a row to use the defaults or not. Creating Causally Independent Distributions Suppose we wish to model the happiness of a third cat. For this cat we want to assume that the causes of its happiness are independent. Here is what our table looks like: Feed Petted SawBirdHappyCat3YesNoYesYesYes0.990.01YesYesNo0.500.50YesNoYes0.400.60NoYesYes0.050.95 Notice that it has fewer rows than before. It has one row for the "all normal" case (this is called the leak case) and one row for each abnormal state of any parent. The distribution is called "causally independent" or CI. MSBN3 requires that all parents of a CI distribution have their Normal state before their Abnormal states. For example the Feed nodes states must be ordered "Yes", "No", not "No","Yes". This code fragment shows the creation of a node with a causally independent distribution, setting its parameters, and then printing the result. ' Create a node for a new cat's happiness. Dim nodeHappyCat3 As MSBN3Lib.Node Set nodeHappyCat3 = modelCat.ModelNodes.Add("HappyCat3", "HappyCat3", _ States:=nodeHappy.States, ParentNodes:=nodeHappy.ParentNodes) nodeHappyCat3.Add deCondCI ' Specify the parameters for the four rows nodeHappyCat3.Dist(0, "Yes") = 0.99 nodeHappyCat3.Dist(0, "No") = 0.01 nodeHappyCat3.Dist(1, "Yes") = 0.5 nodeHappyCat3.Dist(1, "No") = 0.5 nodeHappyCat3.Dist(2, "Yes") = 0.4 nodeHappyCat3.Dist(2, "No") = 0.6 nodeHappyCat3.Dist(3, "Yes") = 0.05 nodeHappyCat3.Dist(3, "No") = 0.95 Debug.Print Debug.Print "Happy Cat 3" For Each anAssign In nodeHappyCat3.Dist.KeyObjects Debug.Print anAssign.Description, For Each aState In nodeHappyCat3.States Debug.Print nodeHappyCat3.Dist(anAssign, aState), Next aState Debug.Print Next anAssign Output: Happy Cat 3 Feed->Yes,Petted->Yes,SawBird->Yes 0.99 0.01 Feed->No,Petted->Yes,SawBird->Yes 0.5 0.5 Feed->Yes,Petted->No,SawBird->Yes 0.4 0.6 Feed->Yes,Petted->Yes,SawBird->No 0.05 0.95 The new node has a CI distribution because of the  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htm" Add method that created it included this: "DistType:=deCondCI". The type of distribution can also be determined (and changed) using the  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist object's  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Type-Dist.htm" Type property. The default distribution type is  HYPERLINK "http://msr-webdev/adapt/msbn3/Enums/DISTENUM.htm" deCondSparse, the type we used above. From Parameters to Probabilities Suppose we want to know probability that Cat 3 will be happy when all the parent nodes are set to "No". The table contains no row for this condition. We could use the general inference (see  HYPERLINK "http://msr-webdev/adapt/msbn3/Overviews/Inference.htm" Inference: an Overview in the MSBN3 manual), but there is a simpler way. We can use the  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/Dist.htm" Dist object's  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Prob.htm" Prob property. For example, this code fragment: Set anAssign = modelCat.CreateAssignment anAssign!Feed = "No" anAssign!Petted = "No" anAssign!SawBird = "No" Debug.Print nodeHappyCat3.Dist.Prob(anAssign, "Yes") will output 0.0099. Indeed, even when a parent assignment is in the table, the  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Prob.htm" Prob property should be used. For example: Set anAssign = modelCat.CreateAssignment anAssign!Feed = "No" anAssign!Petted = "Yes" anAssign!SawBird = "Yes" Debug.Print nodeHappyCat3.Dist.Prob(anAssign, "Yes") has output 0.495 which less than the 0.5 parameter in the table. Why does the probability and the parameter differ? Both account for cat being unhappy because it is not feed. Only the probability, however, accounts for cat possibly being unhappy even when all is well. (This possibility is represented by row #0, the leak case). Properties With MSBN3's property mechanism you can attach strings and numbers to models and nodes. This section illustrates the property mechanism by annotating a simple Visual Basic program. Prerequisite Information The property mechanism is based on MSBN3 collections and maps. See the documentation on  HYPERLINK "http://msr-webdev/adapt/msbn3/Overviews/Collections.htm" Collections and Maps: an Overview for important background information. Starting the Example To start this example, we create a model called aModel with one node called aNode. ' ======================================================================== ' Create a model with one node ' ======================================================================== ' Create a model Dim aMSBN As New MSBN3Lib.MSBN Dim aModel As MSBN3Lib.Model Set aModel = aMSBN.Models.Add 'Create a node Dim aNode As MSBN3Lib.Node Set aNode = aModel.ModelNodes.Add("EngineStart") aNode.States.Add "Working" aNode.States.Add "Broken" Defining Property Types Before a property value can be attached to a model or node, its type must be defined. The definition tells what values the property can have. The five possibilities are: Real String Real Array String Array Enumerated Property types are defined by adding  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/PropertyType.htm" PropertyType objects to the model's  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm" PropertyTypes collection. This is done with the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm" PropertyTypes collection's  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Add-PropertyTypes.htm" Add method. This code shows the definition of 5 property types, one for each possibility. ' ======================================================================== ' Define 5 property types, call the first ptCost ' ======================================================================== ' Define a real-valued property type Dim ptCost As MSBN3Lib.PropertyType Set ptCost = aModel.PropertyTypes.Add("Cost", "What does it cost?", PROPTYPETYPE:=pttReal) ' Define a string-valued property type aModel.PropertyTypes.Add "Category", "What is its category?", PROPTYPETYPE:=pttString ' Define a real-array-valued property type aModel.PropertyTypes.Add "ScoreList", "What are the scores?", PROPTYPETYPE:=pttReal, IsArray:=True ' Define a string-array-valued property type aModel.PropertyTypes.Add "NameList", "What are the names?", PROPTYPETYPE:=pttString, IsArray:=True ' Define a enumerated property type aModel.PropertyTypes.Add "Visibility", "How visible is it?", PROPTYPETYPE:=pttEnumerated, _ EnumValues:="Visible, Semivisible, Invisible" As with other maps and collections, we can use  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm" PropertyTypes's  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Description.htm" Description property to get a quick summary of its contents. Debug.Print aModel.PropertyTypes.Description Output: Cost,Category,ScoreList,NameList,Visibility Attaching Properties to Models and Nodes With some property types defined, property values can be attached to the model and its nodes. This example shows 3 property values being added to the model. ' ======================================================================== ' Add 3 properties on the model and 3 on the node ' ======================================================================== 'Adding three properties to the model aModel.Properties.Set ptCost, 80# aModel.Properties.Set "Category", "Science and Nature" aModel.Properties.Set "ScoreList", Array(2, 3, 5) Property values are added using the model or node object's  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Properties.htm" Properties collection. As the example shows, the  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm" Set method can used to add an association between  HYPERLINK "http://msr-webdev/adapt/msbn3/Objects/PropertyType.htm" PropertyType objects and values. The property type can be specified either via an object (e.g. ptCost) or a string (e.g. "Category") or an integer index into the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm" PropertyTypes collection. The value specified is a OLE variant. In the case of array values, an OLE smart array is used. A summary of the model's current property settings can be found with  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/Description.htm" Description. Debug.Print aModel.Properties.Description Output: Cost->80,Category->Science and Nature,ScoreList->Array(...) The  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm" Set method can also be used to add properties to nodes. 'Adding three properties to the node aNode.Properties.Set 0, 40 aNode.Properties.Set "NameList", Array("Bobcat", "Sparky") aNode.Properties.Set "Visibility", "Semivisible" Debug.Print aNode.Properties.Description Output: Cost->40,NameList->Array(...),Visibility->1 The value of enumerated property types, like the one named "Visibility", can either be an integer or a string. If a string, it must be in the EnumValues list given when the property type was defined. Accessing a property value A property value can be accessed by indexing the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Properties.htm" Properties collection. (For details see:  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Properties.htm" Properties). The value returned by aModel.Properties("ScoreList") is an array. In the example below, an element of this array is accessed with "(1)". ' ======================================================================== ' Access a property value ' ======================================================================== Debug.Print aModel.Properties(ptCost) Debug.Print aModel.Properties("Category") Debug.Print aModel.Properties("ScoreList")(1) Output: 80 Science and Nature 3 Changing a property value A property value can be also be changed using the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Properties.htm" Properties collection's  HYPERLINK "http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm" Set method. ' ======================================================================== ' Change a property value ' ======================================================================== aNode.Properties.Set ptCost, 70# aNode.Properties.Set "Visibility", 0 Debug.Print aNode.Properties.Description Output: Cost->70,NameList->Array(...),Visibility->0 Testing for a property To test if a model or node has a property, use the the  HYPERLINK "http://msr-webdev/adapt/msbn3/Collections/Properties.htm" Properties collection's  HYPERLINK "http://msr-webdev/adapt/msbn3/Properties/ExistingKey.htm" ExistingKey method. Here the node does have the "NameList" property but the model nodes not. ' ======================================================================== ' Assert that node has the "NameList" property but that the model does not ' ======================================================================== Debug.Assert aNode.Properties.ExistingKey("NameList") Debug.Assert Not aModel.Properties.ExistingKey("NameList") Removing a property To remove a property from a model or node has a property, just set its value to Nothing. ' ======================================================================== ' Remove the "Name List" property from the node ' ======================================================================== aNode.Properties.Set ptCost, Nothing Debug.Print aNode.Properties.Description Output: NameList->Array(...),Visibility->0 Events The goal of the MSBN3 events interface is to allow objects (ActiveX controls, VB forms, etc.) to precisely monitor MSBN3 editing with very little additional code. Toward this end, the events correspond to well-defined primitives. For example, there is no "Load Model" event. Rather, loading a model generates a series of primitive events like " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm" ModelNodesAdd(aNode)" and " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/NodeParentNodesAdd.htm" NodeParentNodesAdd(nodeChild, nodeParent)". Moreover, when you receive the " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm" ModelNodesAdd" event, you are promised that node does have a name and description, but does not have any parents or properties. Likewise, when you receive an event that a node is to be removed, you are promised that it does not have any parents or properties. The goal of the MSBN3 events interface is to allow objects (ActiveX controls, VB forms, etc.) to precisely monitor MSBN3 editing with very little additional code. Toward this end, the events correspond to well-defined primitives. For example, there is no "Load Model" event. Rather, loading a model generates a series of primitive events like " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm" ModelNodesAdd(aNode)" and " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/NodeParentNodesAdd.htm" NodeParentNodesAdd(nodeChild, nodeParent)". Moreover, when you receive the " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm" ModelNodesAdd" event, you are promised that node does have a name and description, but does not have any parents or properties. Likewise, when you receive an event that a node is to be removed, you are promised that it does not have any parents or properties. These promises are detailed in the event documentation. With few exceptions, you will receive an event after the editing action has occurred. If an action will destroy an object, then the event is fired before the action so that the object can be queried. Examples of events that fire before actions are " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/ModelNodesRemove.htm" ModelNodesRemove(aNode)" and " HYPERLINK "http://msr-webdev/adapt/msbn3/Events/ModelsRemove.htm" ModelsRemove(aModel)". The events make heavy use of passing objects. In all cases (because of the requirements of Visual Basic) these are of the generic object type "IDispatch *". Summary We presented key components and services of MSBNx, a component-based tool kit for Bayesian network development and inference. We described how the toolkit can be used for diagnostic and troubleshooting support. We described also the use of MSBNx components from other programs. Acknowledgments Feedback, ideas, and methods represented in MSBNx code and functionality were derived from numerous people on the Decision Theory and Adaptive Systems (DTAS) team, and in its descendant teams, the Adaptive Systems & Interaction (ASI), Machine Learning & Applied Statistics (MLAS), and the Data Mining and Exploration (DMX) groups. Contributors include Jack Breese, David Heckerman, Eric Horvitz, David Hovel, Carl Kadie, Chris Meek, and Koos Rommelse. Finally, we have continued to receive feedback from the population of users of MSBN32, our previous tool. We expect to continue to listen to comments about the MSBNx tools. References J. S. Breese and D.E. Heckerman. Topics in Decision-Theoretic Troubleshooting: Repair and Experiment. Microsoft Research Technical Report TR-96-06, 1996. D. E. Heckerman, E. J. Horvitz, and B. N. Nathwani.  HYPERLINK "http://www-ksl.stanford.edu/KSL_Abstracts/KSL-92-66.html" Toward Normative Expert Systems: Part I The Pathfinder Project. Methods of Information in Medicine, 31:90-105, 1992. D. E. Heckerman and J. S. Breese, et al., Troubleshooting Under Uncertainty. Microsoft Research Technical Report TR-94-07, 1994. D. E. Heckerman and J. S. Breese. Causal Independence for Probability Assessment and Inference Using Bayesian Networks. Microsoft Research Technical Report TR-94-08, 1994. E.J. Horvitz, J.S. Breese, and M. Henrion,  HYPERLINK "http://research.microsoft.com/copyright/accept.asp?path=/~horvitz/DT.HTM&pub=2" Decision Theory in Expert Systems and Artificial Intelligence, Journal of Approximate Reasoning, Special Issue on Uncertainty in Artificial Intelligence, 2:247-302, 1988. F.V. Jensen. An Introduction to Bayesian Networks, Springer-Verlag, New York NY, 1996. ISBN 0-387-91502-8. J. Pearl. Probabilistic Reasoning in Intelligent Systems., Morgan Kaufmann, San Mateo CA, 1988. ISBN 0-934613-73-7. D. Poole, et al. Computational Intelligence. Oxford University Press, New York NY, 1998. ISBN 0-19-510270-3. PAGE  PAGE 2  YZ[tu     X Y | ~ uffWh13hd@CJOJQJh13h.N@CJOJQJ"h13hd0J,5<CJOJQJh13hd@CJOJQJh13h.N@h13hd@ h13hU h13hY h13hkh.Nh13hkOJQJh.NOJQJh13h.NOJQJh13hdOJQJh13 h13hd h13h.Nhk"Z[iu  Y +gdn< $a$gdk $a$gdU$a$)gdU$a$*gdU*xl    l~ 9K~+^gdngdngdn x]gd133gdn & Fgdn<   C J ] l |  ).YZdeෲ禟΋jho^Uhdh13hd6 h13h.Nh13h13h.N\ h13\h13hd\ho^hd6 ho^ho^ho^ ho^h13 ho^h.N ho^hd h13hd"h13hd5@CJOJQJ\6e"$]_ DIklq   " 27~L[\v ;BHҿǴҦҗ˛ғ hjhAQhAQhzh*1T h13h*1T h*1T\h13h130J j!h13Ujh13Uh13 h13h.N h13hd h13ho^hho^0J jho^Ujho^Uho^6HJKEI"ty !!!!!!!""""""}%%''T+W+++ ,!,i,n,,,,,,- - -6-;-----..(.K.꿶h13h.N0J8h13hd0J8h%shd6hzh%s h h hjh h13 h13h.Nh*1T h13hdh H+|!!")"O#%h&z&&&&&((2)r))0* & F*gd.Ngd.N & F)gd.Ngdngd gd gdngd&S  & F,gd*1T0*M*!, ------.(.i/0H23/33}5r6778T999_:5; & F9gd & Fgd.Ngd.NK.S.h/i/00#2-2G2H22233/333|5}5q6r677788S9T999^:_:4;5;_;`;~;;;;;)<*<<<??'?K?P?@@+@AAAIBJBBBBBBBBBB̹İ̬Шhzh13hh%s0J jBh%sUjh%sUh%s h13hd h130J8 h%s0J8h13h.N0J8h13hd0J8h%shd0J86B5;`;;;;*<<=e>?'?@+@AABB)C|CCC1DD " & Fgd%s"^" & F" & Fgdz:gd & F9gd & FBBBBCCZD_DuDvDDD4E5E@EEEEETFaFFFFGGHHHHHHHIIKK6Lɪzlho^KHPJ^JmH sH  hzh.NKHPJ^JmH sH hzKHPJ^JmH sH  hzhzKHPJ^JmH sH  hzh%sKHPJ^JmH sH hdKHPJ^JmH sH h%sKHPJ^JmH sH  hzh13KHPJ^JmH sH  hzhdKHPJ^JmH sH  h13hd$DD5E@EuE}EETFmFFFuGGGHHVI{II)J[JJJJK"^" & F" & F " & Fgd%sKKbMMNNOZOpOQ2QQQWRRSS]TUVVWXxY & Fgd%s & F;gd" & F^gd.N" & F "^gdo^ " & Fgdo^6L7L>LaMbMMMMMMMNNeNjNNNNXOZO^OcOpOuOOOPPPPQȹȧȏ~leae[R[R[R[Rh13hd0J8 h130J8h13 h13hd#hzhd5KHPJ^JmH sH  hzh13KHPJ^JmH sH hzh13KHaJhzhd0J KHaJ#jchzhdKHUaJjhzhdKHUaJhzhdKHaJ hzhdKHPJ^JmH sH ho^KHPJ^JmH sH h*1TKHPJ^JmH sH Q(Q-Q2Q3QqQrQQQQQRR&R'RWRXRRRRRRS)S.S\T]TTTUUUVVVWWWWXXwYxYYYYYgZhZl[m[[[[[[ö̢̢̢̢̢̢̢̢̩̞̩̩̩̓̓jh13hdUjh13hdUhz h13h.Nh13h%shd0J8jh%shd0J8Uh13hd0J8 h13hdh13hdOJQJ^J!jh13hdOJQJU^J h130J=h13hd0J=5xYYhZZZ,[m[[[[\j]^^o^_c__`acdne;fghhhijk & F & F[[[x\}\\^^^o^____`_a_b___``aaccddmene:f;fgghhhjjnn:o;omo}ooopȻȪ١١١١١١١١攇zh%shd0J7CJaJh%sh13^JmH sH h%shd^JmH sH h13h.N0J8!jh13hdOJQJU^Jh13hdOJQJ^J!jh13hdOJQJU^Jh13hd0J8h13 h13hd h13h.Njh13hdU+kkln;omooqrrGsst uCuuHvvxExy{1{e{{6 & F.gd%s & F-gd%sgd%s6gd"x] & F ppppppp qqr#r7rrrrsssBtDtYt`t uCuRuSudusuuuuuv;;ͷͬ{m[m{m[#h%shd0J55CJOJQJaJh%shd0J75CJaJh%shd5\h%shd\hzhdaJh13hd^Jh13hd56\] h13hdh13 h%shdh%shd56\]aJh%shdaJ h%shd0J5CJOJQJaJh%shd0J7CJaJh%shd5\aJ vHv|vvvw ww"wxEx|xxxxy0{1{e{{{{{{||'|u||}#}}}ܿܲܿܿ܏vkvkah13hd5\hzhd0J7CJhzhd0J75CJh13hdOJQJ^J h13hdhzhd56\]aJhzhd0J7CJaJh13hdOJQJ^J h%shd0J5CJOJQJaJh%shd5\aJh%shdaJhzhzaJhzhdaJh%shd5!{{|'|u|}}~:?MU"$-DIfM " & F#-DM 6gd6 & F.gd%s +/:Q^}:>?LMTUV%&܁݁-.QXZnoϻ졔vvvvvv h%shd0J5CJOJQJaJh%shd0J7CJaJh13hdOJQJ^Jh13hdCJ h13hdOJQJ^JmH sH &h13hd5OJQJ\^JmH sH h13hd^Jh13hd6>*]h13hd0J5 h13hdh13hd56\]+UVdlLCCC 6$Ifgdkd.}$$IfF 'R  06    34ab p%riii 6$Ifgdkd;~$$IfF 'R 06    34ab %&=Mriii 6$Ifgdkd~$$IfF 'R 06    34ab ܁riii 6$Ifgdkd$$IfF 'R 06    34ab ܁݁-riii 6$Ifgdkd$$IfF 'R 06    34ab -.4<riii 6$IfgdkdC$$IfF 'R 06    34ab JՆ}`rpkkpk]]]] " & F$-DM 6gdkd$$IfF 'R 06    34ab  JՆrĉ+Yފ‹؋ 7:Q'Oyz׍؍𩜏{u{uo{o{o{f[[jh13hdUh13hd0J= h:q0J8 h%s0J8h13hd0J8hzhdmH sH hzhd^JmH sH hzhd0J7CJ^Jhzhd0J7CJh13h13hdOJQJ^J h%shd0J5CJOJQJ^Jh%shd0J7CJ^Jh13hdOJQJ^J h13hdh%shdaJ#`r+ފ['ImՏK & F/gd:q<gd;gd " & F%-DM 6gd " & F$-DM EFGHIqrŎƎ !"deklďŏʏˏ׏؏/0vwِؐ;<@AMNӵӬh13hd0J=!jǂh13hdOJQJU^Jh13hdOJQJ^J!jh13hdOJQJU^Jh13hd0J8 h13hdjh13hdUDLMWXvwȒɒHIb)AbcιιΉoboh13hdOJQJ^J!jh13hdOJQJU^Jh13hd0J=h13hdmH nH u$hzhd0J OJQJ^JmH sH  hzhdOJQJ^JmH sH )jhzhdOJQJU^JmH sH hzhd0J7CJh13hdOJQJ^J h13h.N h13hdjh13hdU$b͕)AӖ2O˗_n!1hIš6gd1gdzgdzgdz1"ȖӖ2Oz{˗+,01ƘΘ^_gh™Ù׼qqffqqh:qhd0J5CJ$h:qhd0J OJQJ^JmH sH  h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH h:qhd0J7CJ h13hdh13hdmH nH uh13hd0J0OJQJ^Jh13hd0J=!jh13hdOJQJU^Jh13hd0J OJQJ^J(_nrskl!156xy}~֞מ !h"ŴšږږŴšŴšŴšŴšŴšŴšh:qhd0J5CJ$h:qhd0J OJQJ^JmH sH  h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH h:qhd0J7CJ h13hdh13hdmH nH uhd0J7CJ;"#hiIġš0178\].Dst긭ꭡꭅynnjh13hdUh13h.NmH nH uh:q0J7CJh13hdOJQJ^J h13hdh13hdmH nH uh:qhd0J7CJh:qOJQJ^JmH sH $h:qhd0J OJQJ^JmH sH  h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH +šDt~\y~Χڨnߩ ɬa֭6 & F0gd:q6gd<gd1"TUZ[op¥å bckl|}¦ʦ˦&'+,89vw|}~Χij h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH h:qhd0J7CJh13hdOJQJ^Jh13hd^Jh13hd0J ^J h13hdjh13hdUjh13hdU4ߩ֪תݪު ͫΫ߫VW_`vwɬa֭® LM_`Ͱհ 12<ةةةh13h.N0J0mH nH uh13hd0J0mH nH u h:qhdOJQJ^JmH sH  h13hdh13hdmH nH uh:qhd0J7CJ)jh:qhdOJQJU^JmH sH $h:qhd0J OJQJ^JmH sH 7®`Ͱհ245=ѳQYҹs,Q<gd"16gd<=45=WXѳQYpqѵҵ9:xy01qrҷӷPQܸݸ껯h13h.N0J0mH nH uh13hd0J0mH nH u h13hdh13hdmH nH uh:qhd0J7CJ$h:qhd0J OJQJ^JmH sH  h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH 7ݸSTѹҹ67FGEFJKs+,<=Q3ŰŰŰŰŰh13h.NmH nH u$h:qhd0J OJQJ^JmH sH  h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH h:qhd0J7CJ h13hdh13h.N0J0mH nH uh13hdmH nH uh13hd0J0mH nH u234NOlmھ۾+,GHab|}BC  'ŰŰy$h:qhd0J0OJQJ^JmH sH $h:qhd0J OJQJ^JmH sH  h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH h:qhd0J7CJ h13hdh13h.N0J0mH nH uh13hd0J0mH nH uh13hdmH nH u+  +,>?IJZ[fgopz{||||||| h13hdCJOJQJ^JaJh13hdCJ h13hdOJQJ^JmH sH $h13hd0J2OJQJ^JmH sH  h13h.NOJQJ^JmH sH jh13hdUjh13hdUh13hdOJQJ^J h13hdh:qh.N0J7CJ0 "$$Ifa$6gd E????6 "$$Ifa$$Ifkd$$If4r @ ~p   R0634abf4'kd$$If4ֈ @ ~Ap L0634abf4 "$$Ifa$ 6$Ifgd$If "$$Ifa$ 0''' "$$Ifa$kd$$If4ֈ @ ~Ap   RL0634abf4  6$Ifgd$If $'+0''' "$$Ifa$kd$$If4ֈ @ ~Ap   RL0634abf4+,5> 6$Ifgd$If>?CFI0''' "$$Ifa$kd$$If4ֈ @ ~Ap   RL0634abf4IJRZ 6$Ifgd$IfZ[^bf0''' "$$Ifa$kd$$If4ֈ @ ~Ap   RL0634abf4fgko 6$Ifgd$Ifopswz0''' "$$Ifa$kd$$If4ֈ @ ~Ap   RL0634abf4z{ 6$Ifgd$If0''' "$$Ifa$kd$$If4ֈ @ ~Ap   RL0634abf4vw1:}(L KLPQ[\uhh!hd0J7CJaJh13hdmH nH uh:qhd0J26]^JaJh:qhd0J ^JaJh:qhdaJjh:qhdUaJh:qhd0J7CJaJh13hdOJQJ^J h13hd h13hdCJOJQJ^JaJ h13hdOJQJ^JmH sH h13hdCJ( 6$Ifgd$If0''' "$$Ifa$kd$$If4ֈ @ ~Ap   RL0634abf4 6$Ifgd$If\w}0+)+6gdkd)$$If4ֈ @ ~Ap   RL0634abf4}[(L /gL-57J "$$Ifa$"16$a$gd!6gd QR\]YZ^_`ars^h fgqrOȶȶȶȶȞh!hd0J7CJh13hdOJQJ^Jh13hd0J ^Jjh13hdU h13hX h13hdh!hd0J7CJaJh!hd0J ^JaJh!hdaJjh!hdUaJ5OPL-5|} PQ}~ 67껯s`$h13hd0J2OJQJ^JmH sH  h13h.NOJQJ^JmH sH #h13h.N0J0CJaJmH nH u#h13hd0J0CJaJmH nH u h13hdh13hdmH nH uh!hd0J7CJ$h!hd0J OJQJ^JmH sH  h!hdOJQJ^JmH sH )jh!hdOJQJU^JmH sH !$%01;<FGQR\]ghqr|}±ﱪﱪﱪﱪﱪﱪﱪﱪªh13hd0J ^Jjh13hdUh13hd0J0^J h13hd h13hdCJOJQJ^JaJh13hdCJ$h13hd0J2OJQJ^JmH sH  h13h.NOJQJ^JmH sH  h13hdOJQJ^JmH sH 6 "$$Ifa$E????6 "$$Ifa$$Ifkd6$$If4r @ ~p   R0634abf4'kd<$$If4ֈ @ ~Ap L0634abf4 "$$Ifa$ 6$Ifgd$If "$$Ifa$0''' "$$Ifa$kdH $$If4ֈ @ ~Ap   RL0634abf4 6$Ifgd$If0''' "$$Ifa$kdU!$$If4ֈ @ ~Ap   RL0634abf4  6$Ifgd$If0''' "$$Ifa$kdb"$$If4ֈ @ ~Ap   RL0634abf4$ 6$Ifgd$If$%(,00''' "$$Ifa$kdo#$$If4ֈ @ ~Ap   RL0634abf4016; 6$Ifgd$If;<?CF0''' "$$Ifa$kd|$$$If4ֈ @ ~Ap   RL0634abf4FGLQ 6$Ifgd$IfQRUX\0''' "$$Ifa$kd%$$If4ֈ @ ~Ap   RL0634abf4\]bg 6$Ifgd$Ifghknq0''' "$$Ifa$kd&$$If4ֈ @ ~Ap   RL0634abf4qrw| 6$Ifgd$If|}n@0++)16gdkd'$$If4ֈ @ ~Ap   RL0634abf4-.uv@DE˶raN$h13hd0J2OJQJ^JmH sH  h13h.NOJQJ^JmH sH h13hdOJQJ^J$h:qhd0J0OJQJ^JmH sH $h:qhd0J OJQJ^JmH sH  h:qhdOJQJ^JmH sH )jh:qhdOJQJU^JmH sH h:qhd0J7CJh13hdmH nH uh13hd0J ^J h13hdjh13hdU@ "$$Ifa$6gd"  "#./9:EFPQR{./YZ±ﱪﱪﱪﱪ¦~sh"hd0J7CJh13h.N0J0mH nH uh13hd0J0mH nH uh13hdmH nH uh! h13hd h13hdCJOJQJ^JaJh13hdCJ$h13hd0J2OJQJ^JmH sH  h13h.NOJQJ^JmH sH  h13hdOJQJ^JmH sH ,@::::$Ifkd($$If4r^) H4J 0 634abf4"kd)$$If4ֈ^) H0 634abf4 "$$Ifa$ 6$Ifgd$If "$$Ifa$ +"" "$$Ifa$kd*$$If4ֈ^) H40 634abf4" 6$Ifgd$If "$$Ifa$"#'*+"" "$$Ifa$kd+$$If4ֈ^) H40 634abf4*./49 6$Ifgd$If "$$Ifa$9:=A+"" "$$Ifa$kd-$$If4ֈ^) H40 634abf4AEFKP 6$Ifgd$If "$$Ifa$PQR+&&&6gdkd.$$If4ֈ^) H40 634abf4{N<GeOg(" & F(p-DM ^p<gd"16gd !MN12nost~!"껴ꚻ껎jh13hdUh13hdmH nH uh:qh:q^JmH sH h13hdOJQJ^J h13hdh"hd0J7CJ$h"hd0J OJQJ^JmH sH  h"hdOJQJ^JmH sH )jh"hdOJQJU^JmH sH 3"&'N;<mnBH^ceNO@eøtatatUHthXhd^JmH sH h13h.NmH nH u$h"hd0J0OJQJ^JmH sH h"hd0J7CJ$hzhd0J OJQJ^JmH sH  hzhdOJQJ^JmH sH )jhzhdOJQJU^JmH sH hzhd0J7CJh13hdOJQJ^J h13h.Nh13hdmH nH u h13hdjh13hdUh13hd0J ^J(5@Yl} B 3H6gd1"" & F(p-DM ^pef()LMYkl>?JK}B}껯껯h"hd0J7CJaJh13hd0J0mH nH u h13hdh13h.NmH nH uh13hdmH nH uh"hd0J7CJ$h"hd0J OJQJ^JmH sH  h"hdOJQJ^JmH sH )jh"hdOJQJU^JmH sH .}~>?BCrs[\ ef2378黰k)jhHhdOJQJU^JmH sH hHhd0J7CJh13hd0J0mH nH uh13hdmH nH uh13hd0J ^Jjh13hdU h13hdh"hd0J0^JaJh"hd0J7CJaJh"hd0J ^JaJh"hdaJjh"hdUaJ'8~pz=>HIhڼ}p[J[7[$h"hd0J OJQJ^JmH sH  h"hdOJQJ^JmH sH )jh"hdOJQJU^JmH sH h13hdOJQJ^J$h"hd0J2OJQJ^JmH sH h"hd0J7CJh13hd0J0mH nH u h13hdh13hdmH nH uhHhd0J7CJ$hHhd0J OJQJ^JmH sH )jhHhdOJQJU^JmH sH  hHhdOJQJ^JmH sH hiBFHDEOP^_%&=tu(){  E 껨껜껜h13hd0J0mH nH u h13hdh13hdmH nH u$h"hd0J0OJQJ^JmH sH h"hd0J7CJ$h"hd0J OJQJ^JmH sH  h"hdOJQJ^JmH sH )jh"hdOJQJU^JmH sH 5&={  N v ~   M"}gd!;gd<gd6gdW1gdW"gdWgdW6gd1"E L N v ~      M N [ \ ] b j k         GHUV2Ϩxxxmmbmh13hd0J ^Jjh13hdUh"hd0J0^JaJh"hd0J ^JaJh"hdaJjh"hdUaJh"hd0J7CJaJh13h.NmH nH uh13hd0J0mH nH u h13hdh13hdmH nH uh"hd0J7CJ$h"hd0J0OJQJ^JmH sH &QRdefoq{!"cdtuv{|}ȿȲȲزȲزѧѐ{hH\]aJhHh13\]aJhHhd\]aJh13h.NmH sH h13hdmH sH jh13hdU^Jh13h.N^Jh13hd^J h13h.Nh13hd0J ^Jjh13hdUh13hd0J0^J h13hd0VXuznp'(>?~ʾ}vkvkvkvjhjhdU hjhdhOhj hjhjhH h13h.Nh!h13 h13hdh!h.N5CJaJh!hd5CJaJhHhH\]aJh.N\]aJhH\]aJhHh13\]aJhHh.N\]aJhHhd\]aJ'(6luvw &`#$gdkh]hgdX &`#$gdX$d-DM [$\$a$gd.N$dd-DM [$\$a$$d-DM [$\$a$$d-DM [$\$a$gdjgd!.5 jk"$H!bcdefghuj6hh8 <CJUj5hh8 <CJUj\5hh8 <CJUj3hh8 <CJUj1hh8 <CJUj./hh8 <CJUjhjhdUhAQhj hjhj hjh.N hjhdhjhd6(hijklmstuwx~¼¼hdh8 hkhXh@0JmHnHu hX0JjhX0JUh h0Jjh0JU hjhdj8hh8 <CJUj7hh8 <CJUjB7hh8 <CJU$a$$d-DM [$\$a$gd.N$a$) 0&P:pX/ =!8"8#$%/ 0&P/ =!8"8#$%0 `'!DyK +http://research.microsoft.com/adapt/MSBNx/yK Vhttp://research.microsoft.com/adapt/MSBNx/!DyK +http://research.microsoft.com/adapt/MSBNx/yK Vhttp://research.microsoft.com/adapt/MSBNx/!DyK +http://research.microsoft.com/adapt/MSBNx/yK Vhttp://research.microsoft.com/adapt/MSBNx/UDyK 8http://research.microsoft.com/~dmax/WinMine/Tooldoc.htmyK phttp://research.microsoft.com/~dmax/WinMine/Tooldoc.htm:Dd6 X  S 4Amainwind3R#kgJԧjFb#kgJԧJFIF``C  !"$"$Ca" }!1AQa"q2#BR$3br %&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz w!1AQaq"2B #3Rbr $4%&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz ?־(xMwA5:s@(BN3OAU6?W[^oi:zt:@!a30 6c8xU,-?iVJ+5V[ qNHCV쵿߱OM/ ZDʹ<KUd¯?MUk渷#kpMĖ(se 1}4>Ie4D'Sa-KUe.uToWeѱ_D,-o?*5} ?Hm *RF,@-!O#Sf5qV2i3VF*3WK?N)rw&I[69uG?6 #UI*[OghU v0qH+.Ó6gWneY ! ^B?w88Ch-cv A8sYas+SO3ӑF+Z& _D,hZq^I sDICC Gj4]UL ~eV(l %@$Eb0<ˑo[-yF46?WI:\l߳~gڼٌݾfmٜ|V;RNK)"Z8bp:VXWpti+[6IyVw6?WIjOQgբ6mw߂$U8ۓ䊆OIGaj/7;~u]޹PIq26׽mvVC7Z& 6?WOmِi~!vJ.2zUoxf}٤t[{kk$Aaߑ *Y]VdM]hףO!|_MQ+Xjfo/E?6?W\j;CR:,.m?*.m?*)%37tkh+? h+tU," wmqTgwmqUQTT3;T4t"ty]D@dU]D@\Usf#& & ⫝̸U,c7? h,? h+rK.ͫ~6?WF~6?Y\K-؆t]D@\U]D@\UsUC&~6?WF~6?W\WfϹn]D@\U]D@\UstSYV!ןs3QT7gMNIMQMW7ER0ݟfuWS7|.m?*HjO6t6?WFMW6hK&y]Nw& pfΓw& Y&gt6?YFMW7MdxN!渞IwmqTn]D@\Us]jE7{IwmqTn]D@dUscCs]7|.m?*?R0}f_u&MQwmqUU?}IwmqTn]D@\UstSW]fgu&MV:f&՜HAXAEp׭ׅJl e^y/9FSoQAx"ygm uM1ʫ`vfWjvo6}?Y|5tln<ӭx^>{߄?תzw,-iv7Y<2誯6 Ke]Aw/Z W^W#˿BssKUmxz=ٖ#̪i7"Oo?EO^)ec[hPͳX]عd8c@Ѱ'{ ҼMi^%Fmm9B0VBwrvS+лg<B=5ŘfY<RN.I{5gˈiJVn裏{nXU}U%a;t96t$ҵ5oj].5l 0|0T~jW z?^x/O{kb=H]ZM[mzo,✟3%v^ݢ]I5ʸei@Njx&Mj]je&mud@Y H+0ø_<B=4¼_ ֟[$֚NqmJNI}~6' Ķ6?d4\]\LU|Ͳ 0[yki*Lu6]]D16H;c2H^x+?{hyL+MN1S5>guGMKǫߗ|it+Yt).nsZͭhMN4KX6-H 3} _.G+нitψژXlݜew-Zi;][xG,' kbH8Aq,6 )CAD[ ' J7W1wy 8H<灁<B=4¼W י个֬*7~oy/})[\KQx{A缷 Dwg!8מOJ4xXzGyD"rۘ(~OW z?^x+?{k3Y:sjkKI;ݬDSgxJ V6^E7ٮ%~QF0y^EQ WX0VrFpz_M¼W  _.]*T>k[h=K!>Uh^x+?{hylYq%/%֓>UGs_U¼W  _.Ow$d:}?ʔW<B=4¼W ܿ̇)>|i+W v?^x+?{i8/ MJ+W v?^x+?{j׬ܿ̇ z+W v?^x+?{j`d:S}򡢾yl炿]ǿƚd:Gʂ炿]ǿƏW v?xZ>T}W _.G+лgR>QGtW՟<B=4¼W ^ ?!*>)}Y _.G+лg?X?yu^)޾yl炿]ǿƫ"}37}W>Syl炿]ǿƫ"&}%2^SU_)yl炿]ǿƚ/S?̇i~?|E}Y _.G+лgR/Ṡi~?|i }[ _.G+лg?KdRWտ<B=4»W ϩC̿>Rwl߂]jyKd>#/Ɗ炿]ǿƏW~ vk̻}O%2 m~?|ږwl&?]+?{j'e/_Huwl炿]ǿƟD}O%2 _)?W v?^x+?{j(e/KO^x+?{hwlG.3~%?|E}[ _.G+лg7R˿_!n%?|޷ >)ΑY^Iho|D5+imtFִYj4hm GP9Bz(Ye]Iƒ~_ѳЭCGޗ.x-'Sn|Ax]UE# G5j_隟OEaHnN >A\¶tNl4"#7l# 26 T~ . KPWѭ.Kw ;h`WB6Zxq< 'tO\[4eqi܀6|pp2,A kЭ|Om|dt[ɴbAɻ#=~ꁀgt7S&Ӽ9 /,~Y7uY[ ּE$4o:ᔁ ]-BV}iiggy-a ,O?7<1Eas˩tye^oi:z^h#L_. (,((((((((((((((((((((((((((((((((((((('znaO$׋oȉK2c8^⟇^|Kp^5K/ OfH-b{\ lpctc>>z};HGv5oson vP=ln>.̪nqGQ_7 to zoR:ʹyΝnȁbA;T$? |-KÚt,ֺƳ-͸v1HI8*A2+DN|EOxr,|CGxĩ2@`ʰL^G gڷ,.k]#M mz=ߕ'-IN@<d8dw>G?j|0u  >iVuO ClfIX\=]gNMkS$(ݷG-OQo.Y7}?[R/|*X#'W3E"ك m?j#yn"O o 갠$T2d2*M S[F'­cGxN{+xmāw[ ³oD&)Ԟ_7j|0taE;|M|.RO NnEsڛ[qN:V xOE%՟%=7j|0uAx{[uT$ RH S#ָ?V xOEs/:-ׂ#\n[[ G/y$ρUʺH+᜾ Л;;M PXYkkodq"*sTjટߨe+iE%6ƂXXޥF!Wde%wg 9#t]ѾG*U噸aIxfa{ݏu:Z-j |--g)G;#V4"/w$iȻʪ\'.2uyh&Is[s(W 㨑)7SiMK_af".Y@9R8#=Ux:r嚳);QREPEPEPEPEPEPEPEPEPEPEPEPEPEP^u^w9GoW?69s|=F}w Šw1cco-{Er)=a-( l[ 9=N0u?w1cco ]Zc1c</|kvP_YbcI"f(nYk^˨u2HVCdsc  $[k?y\A5$I &m^g>!ɰ׶`oo)^6F J4=}/_MB(= ((((((((k}Lx_HG Q=RG=Ǘ*de y01AC}kke=7VS2TaVSR 5ٗ⣄F̗F)+̾K· /CG$Q6Ld%eu mƼTdž.G|ߴIM`D Ӆ\᯲GLm$>"۳c /$@`K|Lqv_U%+Х;ic(+[,O$+_Т̙ Bb[~%jPbpTHwx$`fqUW|SlCFe}\\ŸJV\d25xZQo0椂Pj7o<"B(1/w&oJg}wY.x5 = cK'h"0=j|k).<3qG嶝.d`euEL@;rEtӚ̖RmSڲk^kv5mEİ۽Fx.6Plca}Ck[ mNg=͍F m̳C*OmcY-j2^ŧڡM_ʅC+nEU9PֵuFJ-5ya]+\18% d7 1Zao3wcЭa=CKX!]-Awd%HD6&-"d(lK̞2\gN&4QHd9l>ahUE+;awuvmJc/Vi$g]D01,Yv u剶[͞tz#.XlrYpĚ~_g'i.mJEŵ|2~X/#[x#JV+>h|UE<o쟱]dkk]B9v8y˼|pw|;{hj;3OsqrD#`@07LX1SoiwZޯr1u," #`T >%ݭn/g\[O !U|:'Y ,CRrTekmeg 6FC H#EUUҪ1iYfCfcc}3|'hvm11zƏinG/&%͚A5EtIc9זOo +l.6RO+HmAFrp;Mgׄuaw=nlVYβ% ~;j%IXM6bx/íixn]sUHyKT#UH@Ad?tV汩j:mʹvB4-It[2y49:mj0,fIidlzA rh:Lj4=.M.a6m".`?vq@լaͪxVMc\|7nPҦȷ{w]_2e-?M-E߈ _\\ա9u)1%K#pȮ W;X~ jڌ׾8ibGow$)PۑUNT2ttW_ MrzْiZY&3ݎ: ``T'K%4:+!=RX<x8#GB$ .X*8r H$ Kv~Ko%k4y7+~\3ai_V+GWCyFk<+hcA뺣޼Kn(֧!u'$*'?"O|cI|-hWwgH^f#)B38xKཕZg. ծ?ٵ͍Ti1RXXm?f-OZ j^*UB].d9`Cs7zDּ$iVg7w)HoݬĊub}>sPN{w}((((((((((((((((+εskßn9h>mעםj&ׇ?s}T_QWc?,}lmխsC }Wpn>~? cޟ[k;o k"[*rJC@ LWcWE-;RҠЦ(h'Pp:W771C̓䢋[ )ΟH،[*)Ʒ~tWᯀm: ,R^1rz)\ n!ɰ׶`_l6(gf~[JWzRm`3GO_Kl|P+H(((((((((((+7Z;$ IW\ =pOiQN2qw@q8 -s^KȑؼQrRE-031V b]e!BOUo?̿?IʭaOǏ2'*zf_$VWG3~|};\W.dzYVr4X!yiehYأ)LE*.̰ȥ ?UFψ1C> ԿkEil 0[= C?~Г[=^Esh?'ccǿ~Г[=[#A{; ѼBLϽ>Pqd2⦭*6q8 -mml,5!@Ċ08 YmQP0(((((((((((((((((:M廜?Q^^u^w9GoS~g?E_}ve=t[4%c< }m9Rr1j~? c4%}Yhc~#5olL9}ckPXi]a c?hoÚ>}vnݱFnlgO LL5MS𯃵.hI%\Li%,(eX?n-lÌ52HeѰ%@b(ڥE/]=?M.[^;2:EUyHE…+lmPT${&g=ҴX] -|1v@RTo(i?_Z!co`ӵ; NyI'iܰbE. U 7ZFO9QYOH$x + ö_?fХ<3 m#K6TԻ~jvWE?,_,k}q30o/Xs.neDi-eBmq$}?.vT67VP_XukqxFVV SV-XaEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEP^u^w9GoW?69s|=F}w Šw1cco_x$#v$6QPը1j9=? y\|C%#4m=!19i76Y ҹ豈$mmS dz\5ۭLi[ Hfh᳝ Fm<f%l;E#;#Hqv%8WUx×6ViZ}V8Sn7oFڵ}# &m^g>!ɰ׶`azxg<-%+?t|˟{m6 (,((((((((+[@;hVR@cSoȖD Y W s®; +aöˍZ]A-@u`ǐc|Goj(;Z.r֡.6 $ %͏xþ:͝[>Hby,"bY@+ ST<+b< Z6mcqW/~0nKalxo¾?揢oc;wlQ[sR58>[? ~eЭxqFI6 B+_~g%7'6kG.cqmTŠ>5 +Q@Q@Q@Q@Q@ {>Z͠js{NrM1;xN.Yf>[?B>fmy񅺵EH 0F$0,=WV8MuoL|vjNHd00]ޛZ''\JB@l2uX>"wuu-&JݚVX`B3k_joQ\#0wǁ!XH$mעםj&ׇ?s}T_QWc?,}lmw Kxal5$$[^rA3l~叻퍿÷XuM;Bu0dcNP2I Wcxx+WP_gkԯ2v暗.aHPoBJ(-HV aiwDxKtekK=N-TyeJ,H4ʿq%;R$ng M|Bam3DEK:0 =_Rm`3GO_Koi:z^?ekQ_zAEPEPEPEPEPEY8d`LlDΡ ʌsc\lf--ylCe3$X_?xQFt#6 qeI&`<3F9igonzd֜?&xWxzm㒮&_aЖpÔ?{|l7 B[a$B2 h1 KQR"XZ> ?)<7\9s,hph vQQ:QE ( ( ( ( ( ( ( ( ( ( (1I_ xxsG־ͻбݍw۝uCM/u,r-5iVi¥ *ӗQZF⬞_sVG/Z;m༱]>N I>>`[^ 꺜ZT:5Y4J[ɔ)bop#uR+ ZVΛgXMͶfL0ap@##( rωIO oguGH⻎ddF BSFOcKO x}Gu֡%"Kx\°K+csKXJ[?BaEV# ( ( ( ( ( ( ( ( ( ( ( ( \[=ϯuZs{_6G=pcU1wo/zmݮuae%ekx&򦉊<+hb`Bc?,}lmw/u_>'V%ErA''y>RLN| gl#GŸxsPֵ--RH4y#fY 3G#I%g qmy:xsN0?ׯ#+_>4R&b i˿o;__ M|Bam3DE |/ J4=}/_4x[JWz/?mQE|YT7VS_\kkoK<8HEff<IBWj+־2(4u ~[K傍DYۓd5_~-째œ^Ilot$^5PǠd@溡Ti.Q]OgkH~.6?W}Z{3]MAr:]ÝX$R@3dݕ'/9WD/ O"jŜ3gI#Wx%qx(#HG5O9NޭG^%Ks_>Z֟e {_:~z-j֦בɥB)FɔɸYѳo-QIƱm~/fA.dpNI  f|DjNi=S$ڵV>!k O=jkZC-ipU"CCK)O,2h (,lm>ƱA1 H *o/_-GkggSVnH7~ɉHb)oe=>,HgKxt$ R='4_!O(TiK.d{-w_5-+OWX].P+1c;Pzoo ?xZ殨{M~3oo ?xZi>_޿/3k|oâ~) |hz yc/%x#$#8|(v Oj^EqĭG֋:OcVV66?/ȯp*wy'Z2l?2iUN=q% }43,j'qeBUI8SWD/ O"YiWk4˿i[oo ?xZD/ O"k4/_4WΓ{x_-mfRhWqjoR+aB) NKd$XM;X!`1s#h~E%ůg|%[cH.*V! Cq'OE~ jZ}]Z{X,2'r@Y_a\ğWX·WW,PAnJp% 9$U\8|J^ETQEQEQEQEQEQEQEWj&ׇ?s}E:M廜?Q_]?Ʃ3~X;uGCVŪ cC*~X;uH縆"N2݈}}c;ud%$Djn@bCFcA7L8XuM> b`*]n8ECIܢZy11w6@qd$kp.Y\¢!9Ntм# cs%->3.B8ʍGTp|=ZkG=>;MG+u"El00xuy SgB'4-Fm>:Rj7&la( ³y^l:m5]n}7Bumbkb-|" S$G|)0$Gӕ/Ck\Loe/#i/4`l6۹&X$v;[] íZKç[Iyll΋zuWl;x]Apv䁚5k+Mo wl_?|ݷ~1޳o6:}ˋm@5{行[GG9/"G&JÔk:w~6|#wCI[jmEdUq?{Z!w_6{5?1<~./#u=4ܥwgC-KX].}VҦHlUWgX,o0X]jϮ/!#4iy yՐ啰oeÛ+mP]Y5rr! #1xOՓ]ZxVj`E]M)]`vMG;K۵L*QGյ`2$b ug}&ҧo>'_WWw->4T:"[7dwewˡ #n#v6j-wþ &t ļHb-%[hmOr%g^Q|k1TZQZU8-)P рJ+Mmw oO'GE9e|:sǪ%i{萯 ̠@ot&g5;]dN:W':@hu_&4+g`fɺ@ *䁴|[n^^Bu"EPHR#R!"DU K66et=rm{Xa(F& #|.!/Mtk[|X.YwpA,GDF8q\u⨾_|62[xr}.M K{&5d9$8U2)9-<^Kk$ ͩ*$.j"ebhG+-N}Z N`Hh"6;FQq -v_.M>&xv]~JC_"hY# o V*^O"[Om'-[O{}ZƗ|ˉ"XVYreVRYF6:vJtkIA7EcmO"ܞ2Lw-!>d׳Jd-bEf+HJɰ[k֧k'JŽ}oc~DעI"$sbh`?9;ĶydFORY]G:C3?~1^j6si)[7 KdT!X3K|%g7jsN#?))A ,͟Jʬ-:?w\kAÌŴ?%puqb`' TmaZi>&%gshڍYXkzPidykKB[ba׵ׂ"}xkFѵXK ʗy| gaeu'卹zʳ`! ={_1 (R((( k?*+i9#4 Xm).MFUu;MzS^4pFo*@\gBpHtZ3d[myd8¤Rv`x+0n ^ƚ^B|Qya67ۑq8n3s@?x=&1vl$M~g9φ6ky\ե *%Qtl̬pA5+[-m$ Dw [սRcz&ٽ7Wlk4P4H'0)ڬNYQݍăY7=[.3u%VAe&UۖPeE,̑g{H ><:|2y|)񯏟H\OZ]?Lb[xJY)e2`z~MB"P˚ f(څQ\ ( ( ( ( ( \[=ϯuZs{_6G=pcU1wo:OaԵO w%}"i]c?,}lm>+Y]ç<\gˊ"̀<6s:N=Kx~WNlzΕ>ç\<(;)µq,땾C{i=w2WVɹ-c #2HqAdk[ M|Bam3DEfC S-%+?t|o)^6F\i__QEg ~knmO~77ٱ,4jpwfT(T 1%`-ږ$-ƣjNNIz+q.Dq߆'[L͵lM$,ok$V\gE@A#miɪ[=q9O9\&v|WC #??r\Q ˹og4F\6H*Gr8%g׸Xύq^x]%ݷa Hlu0rKbPbHg練_ {wq܍+Td}ʊS!Yj/nWO"b1 /]F{uO~ x ƺ۳e B*#q CH>EpϳW~BOs7#h}nkm6ϼs9ǤQEykTRM7qEVc ( ( ( ( ( ( \[=ϯuZs{_6G=pcU1woo;-b+$|OxF%u4@=O~X;uP]-m&hKh>^~n pag+%OSy {P\_IӼ OWڃd`ܧ=*~N_/"q=7>s+zu/ ވLJRXXPjFWo4 .>o' ĤSI2_; +z߅|aܞVYo'f,Dr61'c~*lj߈}k|<ۻc\JQv.(QEQEQEQEQEQEQEQEQEQEQEQ\oZϧi4Ri|lާ. ` D?׿/!巼X@[dI[6A>AlȮ anսttvWO<|f-έ7̵ۜgq;)=xV?lko2($L=/hڜZ]V-jF񙔫;02;-ً6I95UBt[yafz.zdZYVnm'Y Tu$A^𮫩˪ͥUhRn%B:aWf+ʸ/WKm~m$R$>NC;>>Ry)ᕽ_jvW W<y#{ sA2圷G [ixw~ 7tB(̓-‹p %HYp@? c_ 5ȶv!{n2IXL6@˒Z<_ ].i#򠷖w;֮N~ \K]x MC\@8'Cvb$=ƹ Y-ogzԒF|U$%ص4wsk,:ޛڼ$ Brl+u>'=/k?;ߜn[:^_9z𞃣k hM*Ij{{x7[`iw8Fe{O5Tд[x -˗9!.K/˩#uKRoWRHd0 1T[;zi 66ƱA(8FUG@8TQXqQ@Q@Q@f@|Gd>!4b$֩TVSat3wGr>%S7w5o yY'&+r@Oa'4?.vVouvO}oM-RCctƮ%K! 6N:ZҬf0)QEQEQEQEQEQEWj&ׇ?s}E:M廜?Q_]?Ʃ3~X;uOϸ'ηo.Xma_c?,}lm~-a xP8< { RWwGu޵E>lywG;#ݜHĀ Z^rjCe# DPWq. )5֙T&2JMnTc/EQMt^q&m'`؃qr!~bjP4< {i& (&m^guzxg<-%+?t|˟{m6 (,(((((((*caq 7I$Fn2GʊHg$ &ݐ(x.lQXxClVx foC)6SsyX͊tO#;!xLv& pp{9asbǃV/3bNe3}IFœxClVx foC)6Sϳ V<'}y/&GvC.<M6V/3bNe3}IFœ}\بomo짱6x&@ʌ0x*A `>7+حDy2?whlRm'}y/&GvC.<M6T1]˦i'еK.oVF` (? ?hnvcCk9;pwlA}o+W[Ήd'd2əؤ#oN\qbL!FVPxClVx foC)6SsyX͊tO#;!xLv& pqQ|0Z7tOH1~m[yyvnڹю±>7+حDy2?whlRm'}y/&GvC.<M6']6(xw6(x.lQXxClVx foC)6S+,fXdE@dGXIQ5QREPEPEPEPEPEP~5KIq"Rf'lw15; 17_i+(GʖzT56p Dž '1E?gNSyWiw8x'67եˤI-u[ @lq/ZH?#මD.U#U`#* +#0|q9ARu~nS*J`8Rc&gp9(խ!ɰ׶`_l6(Yц_oi:z^h#L_./_/( ( ( ( ( ( ( ( |_nxGDєҕȔ Um9̈_U\ěO}_?f-vؾw,7C}AqtY] cCחT?R{v=:Cjcan"l2[/uK?KhN>رI?}wKZ^ |1{w}w5Is]1 J/}~4|#8~m+MG &#!p`l?'.irkEi_T$;d(IݞI4UH՝;Y?5/fȎXIBr|%}'>d?ddI?}wK^WWGƺγ'm[6 Z\8cir,;ȓO B5eio{~4|#8~m+MG &#!p`l?'.irk ?CO hGgp$>V}MG7$Ux~em.M/Əv;}g-Ci>#}O_g]׽G?=}xǟ5+-2Gl㲎+v[e$XC:&44"Ho)ݍ->;p$>V8On6F$̻ɮX7uF]KWwٱi+fRNgS'5X˃vg"/tZ@i1x0, D#'xdm.MM7T=?dMdW}59|7*wv wx%$I9B]ˇbF8$>Ĥk/%N%H|op`lXF$̻ɣEng{8~m)'7@?'.irk[}";q񥬆7uҚn R $Um?6 g2gFhwiCk[< nW% nw.V΋R'tp$>S8On6,#}O_g]Ѣ7T=?dMdW}5X"?|-H'Լ-v~o1v2!TN ]/<|7v$z.dh_I`6 (oݴj`<\V_nSvQhf>{fK(fxta517rЛx1o m?8<#ӭ򼭘FݞT[qo:lO8in-߲z~cS/ sG_lG%|G߅4 o|=4>iӮ"!xs$0QZ\ N_rZ?_!&}wDm:-^3,rq* e>F>7uti6׳}+i#IdZ9/b;G8?3wxXmT*tT:o7:6 &S;{ˆ  '{VM4ykgx`voo;s3Չ?^ K{rku@I Ԑ2Iǹy>I_A.杏ğWX·WW,PAnJp% 9$U^M|#a]x .75[J0A pFFA\vh"֒I"Or^"Fyr:9q\8z.YEx%vB.RD 5 [DE s# C藱i/[}wFqWCkf,$kp^: iEx:o׆lOv,Y+FЪۥKk? u\xxqm ٕ,J&)v]8mHM#z,|㻱[ Fy+8҂QImbTOO2&m^g>!ɰ׶`7|/ J4=}/_4x[JWz/?mQE|YQ@Q@Q@Q@Q@Q@Q@qm> ohgE̷}by 1e{ $}bϙ5ngٿb "zI(|M >[ۍReIc;0:%L)"Tc߇:5M GǧY*𠐆cot9*3wzEw^ܴ1;d6]t)4&~L. |*}<tdY'i$!FČA'N~)x).~B: +?}|T0[fW-Vٱb$[B 2 h"{K/?5<7Ŝ(C`I^[Wyi;(+(Q1j0mc]ΪWx 8,O?~ Z t{E- +'BO$99<WHՋHEy%ǝJOx/5cBh%YpIF)jባ5x5X~k|smk,679ۜqdg[g(>ͿZE6K5m4+\]JReIq†#C>{x#YHf}cU,K"GB0 Tso& h:}&1co γ햧w7q6 HfNO$fsegGڼ;]wyݍg7uu桨VLĹ]'!Niz ?gE/ՠ^]\Y7n' ~ 㡺[>[τoo|#oSU4?|жa~߱YÍ7>}v.~ht]B?Y O6]>Amמ=]β-c+QƶM751F$ <*4owƪ}?Bz? n#Y`Taea& ApA/&y*K_n- kQhkp(a FJygʪ_1{H=^O[Lq nkcB \2,yRA4+֯%Wo GhhnINX~$E#-[2[O_^tG׎po{hkP[]+JO 7Hc`й`t/ݍo:隿/Ʈ\gqzm$mY-h~HGS| X<)}7LchEџyV#P83pN4`_)+XMVyula{}ZıX0ٜVqs!)_mOt74[>ys3^A&T|G69Wbm^ݟ-!w7<`0yew-hm g*>i#؛+pr6Axڎbr(:xzmo5=:kXf;vt#k.vGIET&%%|:B .#2KmjBX3w 2m H[_x~#ogn;[]7ZFO9QYOH$x טxyAo2K-l`m-qdF g.r>7J*85_(vtW| ׬dImbuKvBL Kr,H [ȧ>%x[qk7R#*m3Hщx-^;-ѤƷ Zkz}:TZDX[Z,.13y=v]oĺ;j5kH|U.`4 {kfh]H^VYIqC~ ˿5G4owƩ^ i%|x215fľּuQ/gH}V@󱴞ݖZy1h07Tkx<'m[c)&&lUKu rmj_h߃?92Q g/]/O^oM4FW_5-gI5U8Р8=+ӯtiiM>$ RU6dV iByV8䕉P|03CS.G/{{? JB>3,) =H'Rzg;s )K^J⛋h#C,Z;b"HQp@๥V)z"zw^#{i4"iP-ͽ,7<ŵBH,+υI"OۋHuFG8JCzj/hQMJ5>j/nWO"b1 <]xqX.0`kq]7/}x Y kpK hC7x# Ξ6𭧁k[`_wsFxTZ&>G{en.4mJwOvXbOSsRQ-r&m^g>!ɰ׶`azxg<-%+?t|˟{m6 (,(((((((ZMm Ξ.Kլ8% ǷR$܆dt&y&uum[bjo0~dL̮~pXMuG7+Ϟ~l옝̓g셼CqWG֗7Ukԯ>{bv¾L>[2Oݟ}q꺣zsc6vLNXWɇsIBcϸZ(PZn?پש^|6?kgd||>d?d-7MuG7+Ϟ~l옝̓g셼CqQcʴWT}Rl~ɉ+ 0n|?v~[o9ƛoW=͏1;ea_&-ϙ' y>h+Bdžjk;hnt$}N$,nlLґ!c+O|G7῞~nbv¾V-ϙ' y>Z+XMkW{\ + X|>d?d-7MkW{\ + X|>d?d-7hSÑ=i#~ sc7K1;ea_+̓g셼Cqi#~ sc7K1;ea_+̓g셼Cq-iTGZo<f_7LNXW$!o1wo<f_7LNXW$!o1w} EU;9֛?پ 6?ktVn|?v~[o9ƛ?پ 6?ktVn|?v~[o9BGNDxǃ-m|?e mk5픷Z0|v9u]QJ繱_;&'o+$!o1wz|if^y͝Va~s;7Ukԯ>{bv¾L>[2Oݟ}wK+BǕif^y͝Va~s;7Ukԯ>{bv¾L>[2Oݟ}wGW*u]QJ繱_;&'o+$!o1wn?پש^|6?kgd||>d?d-7U 0mmuyb^I䱆P\9 [MBa)č|+-ҧiK(^vJ\,X!g?tiQYկ"V (QEQEQEQEQEQEQEQEQEQEQEQEQEQEWj&ׇ?s}E:M廜?Q_]?Ʃ3~X;u]j5+3 # EhF>hAǥvw1ccoXc>/1}}zaXz^8Y5 UE;Qx'tn-vQU.WkMn̯u8_l6( {i& +gv~[JWzRm`3GO_Kl|P+H(((((((((((((((((((((((((((((((((((((+εskßn9h>mעיkW=V,R#.xSG~ݰ~S~g?C~? c<}S[iw7.)`[R Ka0aq.O8/5;{[I|C>/g3{*RU/rK_s {i& (&m^g8oo)^6F J4={;xh6dVnt |؜o}R)ߙc ߄M{6ϰ[<u2 ׷]%~1ߟU9gُcgw:+{O:Ѻ8?-񎿾n0w1[:+{O:Ѻ8?-񎿾n0w1>GuW/ n>uGuNq[}QqӜac>}n(_}?ϪF뎜{O:Ѻ8?}10Q\%~1ߟU9a n>uGuNqca=3Kqc>=:s>}?ϪF뎜Wst{gQEri_~}Tz7\t}%~1ߟU9a?΢-񎿾n0Kqc>=:s>]E[}QqӜai_~}Tz7\t}Ͽ?[:+{O:Ѻ8?-񎿾n0w1>GuW/ n>uGuNq[}QqӜac>}n(_}?ϪF뎜{O:Ѻ8?}10Q\%~1ߟU9a n>uGuNqca=3Kqc>=:s>}?ϪF뎜Wst{gQEri_~}Tz7\t}%~1ߟU9a?΢-񎿾n0Kqc>=:s>]E[}QqӜai_~}Tz7\t}Ͽ?[:+{O:Ѻ8?-񎿾n0w1>GuW/ n>uGuNq[}QqӜac>}n(_}?ϪF뎜{O:Ѻ8?}10Q\%~1ߟU9a n>uGuNqca=3Kqc>=:s>}?ϪF뎜Wst{gQEri_~}Tz7\t}%~1ߟU9a?΢-񎿾n0Kqc>=:s>]E[}QqӜai_~}Tz7\t}Ͽ?[:+{O:Ѻ8?-񎿾n0w1>GuW/ n>uGuNq[}QqӜac>}n(_}?ϪF뎜{O:Ѻ8?}10Q\%~1ߟU9a n>uGuNqca=3Kqc>=:s>}?ϪF뎜Wst{gQEri_~}Tz7\t}%~1ߟU9a?΢-񎿾n0Kqc>=:s>]E[}QqӜai_~}Tz7\t}Ͽ?[:+{O:Ѻ8?-񎿾n0w1>GuW/ n>uGuNq[}QqӜac>}n(_}?ϪF뎜{O:Ѻ8?}10Q\%~1ߟU9dxFγ"´+<?}11}n~0뺅>ֿ5Egv$_+aO>y{kIZFUg| f035ֶ&Z-M4%.7 8Eùf+Vr&3^HVWUm|w7 RsCGj6on~c2^6ڼ&3i}<EQ1̒1ʐ U _ 'tGR?[zj.K='A5ѭ-Q, #&B1R.ܶ3}Y Ydq[ӬeEKfGwsneevǒjaN ~Pp,R)WFT AO[YOű]]4Y,mmmrt ,V7|i"L ݞ=i%b[~?y/6o?LQ M,|/ J4=mm^ "mTK(ͬl\@'˝Rm`3GOYܰxO*6vcghmu}Gd/5n$$FvȮ'+cp b1woW_6w1cco$it1~z/LtM$it1~z/LtMswc?,}lm>? c/I vc$: ^0I vc$: ^0F~X;t}l~叻퍿÷X_@>OItA1ӌa6@>OItA1ӌa6΍߻}v1wo$}Kzcl?$}Kzclw1cco}c>6aIC?q'N1IC?q'N1:7~? cc?,}lmm.:~OâE鎜c m.:~OâE鎜c tnl~叻퍿÷G~X;u $] tğD8a $] tğD8`1wo}v Hݴ?=:q&Hݴ?=:q&9ѻ}c>6w1cco$it1~z/LtM$it1~z/LtMswc?,}lm>? c/I vc$: ^0I vc$: ^0F~X;t}l~叻퍿÷X_@>OItA1ӌa6@>OItA1ӌa6΍߻}v1wo$}Kzcl?$}Kzclw1cco}c>6aIC?q'N1IC?q'N1:7~? cc?,}lmm.:~OâE鎜c m.:~OâE鎜c tnl~叻퍿÷G~X;u $] tğD8a $] tğD8`1wo}v Hݴ?=:q&Hݴ?=:q&9ѻ}c>6w1cco$it1~z/LtM$it1~z/LtMswc?,}lm>? c]mn EM22ﳁ0+ⶹ*FP$ 1ӌa6F~X;t}l~叻퍿÷X_@>OItA1ӌa6@>OItA1ӌa6F~X;t}l~叻퍿÷XOqY8HE'l&8#@ʮ:l鎜c 27>? cc?,}lmm.:~OâE鎜c m.:~OâE鎜c :7~? cc?,}lmm.:~OâE鎜c [v;VdZHđ3m OL{avF~X;t}l~叻퍿÷X_@>OItA1ӌa6Kk-.7mft('P =:q&9}c>6w1cco$%mbX# 㪮8 tM$it1~z/LtMѻ}c>6w1cco;k eemn]N.ȿ$}Kzcl̍߻}v1wo$}Kzcl.5׍A1ӌa6̍}v1wo?q`:|@=:q&Wv ctzclw1cco}c>6fjt.z/LtM>ON124? cc?,}lm݃]^0\}1?E鎜c dil~叻퍿÷G~X;u11ӌa6ڸc>ˠ8`1wo}v3Wv ctzcl?q`:|@=:q&9}c>6w1ccog>ON1jt.z/LtMs#Oc?,}lm>? c\}1?E鎜c ݃]^0F~X;t}l~叻퍿÷Yڸc>ˠ8a11ӌa6̍?}v4~-ex4{NJHG 8Wv ctzcl5?Ki&{} ,HRq`\E/׌,՞Nב0nG%GOI(6?JU9_ 'tGR?[zncL֗y,vm]qd%WQ2[[Z\H$dOGam#' @'P"Z汚8dkQ CV8=ԩ11&xnjCxL״kOpn[,aJf|*1ַZ}FO*侲Y',Q:9U AN 3υ@.tKf܍6,OE:'sj߈_l6(:h?ң٬pJƹOxmbO@#Ln`7Au|#r[pelnE8aW-%+?t}}/S,-5+-dVHAFJa[TqF_o֩M Ef3MS 1J,A*2 l'\Ӵ+ ye8a4r<K)9 G_givֶַEIf`ۚCIvb_vX9lxWBԮe2$I5RSc ҂TC9O;kUuHF]hǸ<6f wUmz;N;(/΍{M k|B;$;vq3n v GgXNPOtn38Btlĩ =M'RbԆu ULb(F9P^9g%Ry5Dwq@X1dNWr]cʏI__js"{kBm8.=oko/TdGit-q1s{tB M2muu(#g;nB Ex9K__[Ϯ +;O/ )edXdWM>O%nVMf5%%Ŵw:BvbSxt 5*E6% NpHvViK:ڤ+D]P4P;>"PVyvilb9FeX$V@4 9KxAkm"M:=gN;pIXĞb[(亩%ľc & ~RDO [WOދaIͳ6V4˄yNWQr~^ҟDG̑uWT/$cQ1t)"fBd?8P{.+vs{ 9UgBayNW0&dž5y6dנ$K3` cl.H>nN/M 1-?j*١'G١'@XEY4?d(4?d( ?f̟Ef̟E`k?bW z u VMIe(),'+Wz ^C+IUTWwten>6yi5ֆ& '@֍(yi![$Cpf1BW9aU/b-{mUck›;g*vTu$OOǩO%尺O; Qii[$ N`|Ԛō'B?!Yi[ɵe)xɉ0p+d no>nV9 &pTl5IJ^K)HUy‘>cxOEj%ȹp`tT~q/$#W3t㊱h!m|P7+Rjh!m|So Wo Helc/Ojm|U+o#!MΩ&v_Kڲ𧇥Ӯl$6-]G.Dܹy>Ht67?PJ:M$[iq>'x<,DI V6*\0IG`Fr H3[miR K#$meyIe]Zrhd&BC*]Ne9V IhZflY^GIFG**2Wamp5Ϫ:g4w][}bB})hFs3=Cwiws^[ ^! Q *@02y8抉nw~OO?EY( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( (9M_c(秂?<Dd'V  S 2AEvalWindR&q О-QJn6Ff&q О-QJJFIF``C  !"$"$C| "^ !"1R#25AQSVsu46BTqvab$%37UW8CrDXtä:!1AQq2a3R"BS#b$Cr ?LfidvxKrU`;X-K*B4PpE)g^mFlE6u7$ 0[.w$d%t`GNu#V/iv|j1#rZM.In>Z%M_l<)AkfOM:M{kLk.N\ݳen%KSŔIy˗< -n + +'q=@4؉<}.{iu|G?ؒۦε_S5u?Vŧ w;Js=wfh/h^bRH)'ߟ'X~y`Fjw!d)*JH .+8miԬlOT9xۦεGJxۦεGJӲ!= gܸ KZH $>B 1[~GqemͷĈX. BBѐA^myމM웶+)=C}{6s|Q6s|QЭ1զ&èmYb=v1b':Y^u۷ncEԤ%+WV{iţ\Ϲ־(OϹ־+ )/ZؑvGvNB┾'$93ϑ8V#c?sc8FAڤ$uK[IU')՝=ocRron:)v:P:XQ3{sO|n>Z)v:*JVܞݦ}ε_S3u?T eOlϹ=Lķk_Ĩ*ʛ %ϹZ%@W[?P=Lķg_e@֫q'7N_S3q2>?FR}~:OϹ־+ʀҮ]?&e,Y$3q27s|WYl~OOo>)~:PZ]lnO\Ig_ħg_Ĩ ~ze>nq'7s|Q7s|Quy>2>Lķg_Ĩe)y>Oi.$3q27s|WW[fO2}㆙8OϹƾ+ʀW]lNOKOo>)~8P^Yl >OZϹƾ+xߦ}5_T֥Y~X3q2zu:e.:P2JO s5_zѿo&`QVz.daG2'is^is^kLx__>-G?dcùÛ ,1m.PV8 IP^ ^{q9ϯPO>NjQO )NG ȃ]XM:ÐgLr+|L[IZ?+f cShםx舵:BT7$/)䴞~++#^~NsK? :mǣFW+r#,s{q7(I۶3aq[^&nw &t)S[hJ H^2T@ UU_“~̘CumoF dSzm.v!GRZIid'#9#Wwe,߁,cT+?v2PiioAn KlN7<9z<']-SLq[2T(IrL`I3#Z݆lZzoX8mŽSK}nzFv-pPZHB{^΁vէNrzݎ]jTTVjI u[ n,<mt"vl߃*e&:NmRmc HVB$cy#YA:DLx *n-Sq9#*ߝzWNekUӺBu4/!N].jOY~RRK܁Ȟx#<=|ej 5bJ#|66o㷑aYtk#iΚz݆Ԫ߁YU~4P}zWNejʾ?IYzWNei~4P}~7~eJ~Z݆UiMYm(&nM+*۲tqrsiW՚ж{RB= s ; =l w6I *jA~;o*^?RvJ_sW֫^CJ,m_W[bK3p[RvJ_sW֩[fK3x>CS׫{Y}ھ)_ZRvJef[T5Մ㶥kl/Y~-JJuP;(:c}6sdyZ:)>%B)V<?GG7oa >%C[ӟ}92`֫x_C3{T>=j?GG7styZ~$XV<?GG767T>gX&Ƿ_ jZcŒΔ6TyI>`֧D.9 ,dhU:SK? ٺRbHݷݍfZO[4P-8*m**=\fxN~0oSstyZjԿ)P#eҿںrXYIf8u A5U°'Ͻ뛗PyŐRBer3V<?GG7jO~ĹU?FZھ1"\d*E넦7̀;8H _FGr-:ɊjdlOkp9ٖ]\Gd%IZ@ (`Jwť|+p1ÅNm_MnAm:B54>!iیi% zH* +~4%B1 HnG!Xbx>[4u+ۭzh+jᯗi=}8rGܮrP/,ݜ}eT%ck "=s!C>rGܮtgH=kPM!";iCD c$z3 ļۡ-mxGr4mi)Rr$dhYh/DH Z}Gn}5%k]-Pa DdakR҆P!/`99֎T[5+Qrn-Py(2V]uI'{Uҥlޯզ|cQ@51#ܘZC9IKRJP ZY454r:HjyvJxBdFe^8pakcz$*{Cضʒ Bێ %C $aA$ktyLL]q Im$2޼(Q HNxÔa6oKs^xS洚mN6^_Ӊ_9cj.iRGZ,-rCkQS= 2 =_e†4Soa,woR{@Tw:[{ǀ^Զq7a\Ϸ)F8ڷmV,x.Wi_ kWY,0fΪHyEPXv3@TvJ1 OZ2=g1w*8ΣDCy+8hOkr#_ѵ}&^u:iyYɊRN' SiNmDNDG9m:ل%aA$ZV'[Cd$%%eIkz"|,3nAռud^\J Rd4V[r;dʹ>尽%uuq{\wN8\"n\#ߦF'7&!qAM0ӵH-]Y'!@JiL::@ꑮ.FeSXIKc)%$a)!Gko^}"MR>Hq!JNA!@ - MnmrUL--mNFo99W>c *,Kq뒹KJ3Q JBP2B@Eq@$)jڜ5*ҶUu vZ#6uIzG{}]n](`U /yq忭RCLƮnVӞ"epB)H&S`jl[J ,Yp}6B)'dLD^Z~{?5yU5%MeržknC;\ly QV\v)JP)@aÅȊeRg29rGܮrjwIvn9۝Jqߌ==ʡ*pE{B}63܏չ]FpE{B}63܏չ]FP"\1z=)WpǢd`Y=>_%:L]ԯQjmR8A$r1P\k1꜏jtaGq3艛̞zt%ڋt[fA U'CT7iyQuE󪺠MOe{*f{apHobHش˞ʓYuҷA]$)+-ekW-y+d#A=CdvUʚnQ}Uʳ^}1u"Jr)[`(N¥ RS ,AvƜݮzaZԙPWFxR#vTBoN8G$RscP{Ӗۼk=PZaܥmLmܢBR@F+581r̗uYiwCIG%i9INVB \<zbKSj,k}M.][3!y}1dX`BK Q+m,R3gڦonO_.8DBm"ZynR8eiBR gIhZu8(Ȗ砼n k|)h:K%4G#+iK %Rܥk =ˮ߹]L7K1nADxR1mҕ$6)JPJRR()Jg0꽫 g0꽨f׻;7xvBLK2`In22@yb\I{hR[^{~Bյ/'+Kյ/'[^{~BqhR[^{~Bյ/'+Kյ/'[^{~BqhR[^{~Bյ/'*5-v);q}P[;;;P哒X}[^{~BqFꕥN=A|?tnZ^=A|?tKFꕥN=A|?tnZ^=A|?ues]v˲S#"oyV3F}z\xr%izӫko_(O.2?mVko_(O:#Ѻizӫko_(O.2?mVko_(O`.Αcik}{{UB U7l̅ lg˹rŮ7l̅ lg˹rT*?Db{JR- VE#<{}PJuJ_1ozۖ;q/# HbcS9%vg7=0NKfe ̃ a) Oe\qo1ZuUu@Z˚TÈ8i=&'^#EoZ5˾wrHRVZ֮[Q(Vut6ɿGjz,'\4ܣgEֽqicDSֶ%"Q<&s{Jyuy_ն ZcxKHco3nrEqN8N8G$RscP{Ӗۼk=PZaܥmLmܢBR@F+%81%y\浺ZF$F>-2ۓ!ҷ؎X.7$-~|zYz4mLHh5)J]ڢRAyMf)KnV:RĸKEsD4!qi͎wIBJR3ctyc\iql&!#,1ʶ)^ r H obnv%Ӹܨ݌aƔ<Vmm3'D H!.Nj.?wxFcogn;N3߃Vք R)@)JP R4a{{Va{{P1JtedEb::B܅z>j9j+j:oqmR^8p PI}ؑ>~2^9Xq7yJ[pI #`kb{21\V!Y_#˴dK5ƿ)w4Oښ[ʄq-xKP2\W"qc@5KSkBl! I SwOI# 2 +%K6tF[_(ڹm`~qkѽ9pnlI;Dl"Ì4WD```f[N47ߵtq!: $O 5rbG-!] QHTMSstm\ qܙ$gE<'FRHے0"dRchKkt]RRTR|R\4hgލ 8?6RpnFb.tGfk-EmN1YhEp,G MqHRHsnadv ;jm+@iw0]N䯎VQe]pNnd]!hyܥΤ PC33dvzի/ܥO.ʗvSȟ˧ҵWhϖg}Whϖg}<7 VSR~Е]d3)v~.ٹDe3$unZDteJKDoz5uvk[&1.rVTS$FQCqOO]t;jZO>WԷW 845PvD:Ph\k{H\ScFiO}#֡An=^.'ЕVvsdr2ҟ/Hakbӧܰ+{|7D뛩Kb3HQ-UqJrII<$Ԇg}<gm.<:[XovPo&!PZeMܙu_iĶ[nCHQs|ޕ-CpP1 v{*)DXv%h^նbajm[QJb'#h?&3[Sp\`ud- S`e9X ͱCfI: `<4”̄Q؞\$zVߣuLP4%:-d"Trd59%r}%/RRTRzHJ zۏhUZT/ҟ/H<7J|#u%|)AtR/ҟ/HFqoPs_bI6 J!K\q`-9ϞaBQmL=I \gF7mWT^=I \gF7mTH:QGڨZUt꽇1VgKa}WnH)@aÅȊeRg29rGܮrjwIvn9۝Jqߌ==ʡ*pE{B}63܏չ]FpE{B}63܏չ]FP"\1z=)WpǢd`Y=>_%:L]ԯQjmR8A$r1P\k1꜏jtaGq3艛̞zt%ڋt[fA U'CT7iyQuE󪺠MOe{*f{apHobHش˞ʓYuҷA]$)+-ekW-y+d#A=CdvUʚnQ}Uʳ^}1u"Jr)[`(N¥ RS~W_1%bљ\pO"kvxovG`UtvLWZZx[)9>ԺsN<`GXc#83fZrWZ-]#@ #Hәmɐi[G,]BRAP\?KIi,}=6g 4[Sp .Q)Pk @&3uJZYJV=)ڴP=WM\-_]XVK/[bKSi|i6;%m%)JʊR^d:E]LKv%q V%+{AXn.`Ir[m;ʋ-ΌFiIZyXȑ^F p3 Yim JRX\T-OiuUoz\FomFD8eF,:h8!V7lIcZ6DM2j%-Ounl?t-7~f6vvN3=>l%ɍ 0ڜCAN$ANO))$؋[Wԗn/m2HnQ+3zɞ0[ _ [t]޿HN:Ĉl&+q,Q}͉Z*$a$aɍ63!jLgK2%CAK)^Is *KM hBʂTS(Xگ`Ѓ֔)@)JP?UXS?UC0 gнͷasP<IRAA 5jl9*ܖ#Mەp11.:6FMȄy 4)% u*AϘ9oQ>d1dj7 )iNj) }mp&pk:џ,8:џ,M7 VSR~Ѱk:џ,8:џ,.<7~]菆]2ZT6nݷ9ifk۬#ȆH*:R`/yk 457in^^LEh<‰9s$Lo[o.8oCq#0piAhcpI Re8JR:HZڴL&[P#i0 xYТS'x(DvwF1&#K5\wP\1H'd$%݊?rׅxKdd'ݞ6޼Ϧ>ɯ'~ʼn)`vW GigvT6>ɡs C>+hq 0Í3 YkБ) 뗽?EI 6(pMR[Ό7h'3yNaޝ[Qnl0ZJUw/*.X^uWT pZLl8C P)s߹Rk2u4[V>\w$%ej%`WClD?^x̒}UJSM1OVt]k1I[_=kb\",i7Tw]!*Jw`G,o+Ex٣M[(a,e@gDmN8G$RscP}Atx-6צ6n7mFqgfYڤ̴\/YֈLE07"5qPx;_H:ev:[9IJqd\,w8L\dFHSR/(Nx ú}goeޓϺɌì4rKVLvJOp4[Τ̪ۨ΢ %a~4_ (^0/g R@=zݼ*KDݫV#jy RӸ-RXZoXH[ 2)M4m) )T$v$ƅ $53 מuam R$XQasEIm Q#rЦROgs\r5$S}U}_n{tʶikn? $ yrHrGD"OJW^|Jݭ" ԛ3>]P- RTȥDkIh5 ֙ExY!vȮ))Ka+qnI5=ad8c>ښy{hq RTA X%fvK\QC q( HD.9x A$xNtɷ4 H -ZJFӑb梾;K[ )@2#FsF[s!q܀@mݴEpo,O%tԌD@SɌCL%댅!eA*),Wk֦.S פ>fԨny˂8WY%א}rm]IZ6\ַE=bS`]9g92:Z_FK q-L%k}ĩv {[.vۣe$ܮ;q#ry- TEeXی~x{er!m[VC\#҂ jWIe)RTR()JJRR1~3 ͮڰ~3 ͮچaWw6zAVXV<-أ8iŝOqD=ղpNӶ%bӶyW[}J;ₔ6VAJsO%]5mu yqw7(]DWR0 P9n[mmM}G-J MaE$=8r3;Yj', ;ɔ_㩖ߔ-b)\˘WZzPlA3*{ ,-'!H((sj:Dh D j.,\BOiHZN@ s.9-+ЦRTIXh8 *z#neF*Kα *V6N#r-ֻ]ݦnhp܈PTRTv8T dvzի/ܥO.ʗvSȟyӗ6퐬}Ń#w33塧w'j [r#uΫWSBܫHDb>p1P|`9lzdC\۸nJM ) QH+V)JGJRrdʸqz7~k-!ɲQ۲AGdrajD\Mޤ$?$=m5$8enYҬ iТFm| I8*H8+iRA7';uFru d teҶؘA »h:9PxKKg2˒+JV-p2#{Jz͡;ѻj",}A+}rRu{7fVȋ`oH;V2WmMXbwb7UbLom.Rҝ̈́ՁoRl$YWhϖg}e>sD|0镺QwK6]gD7zGYTMΡԺ$|{XLi_QI#Op.ޣpx:gxr+@'Nq˙"c||֑uEKzNjKJ F׃`ߴ;I( YjGZ QbVGNY l۳=;ם5/ر?,c8]b ᶨ,ʒFԞڹ4.at9tgu:!Cf8qFa-v0!8r!ҕ痽?E2蠱Pn;;B".$Iң)+AW,J$xkoZ˵1d]35T#?j)Emj&ׯ5;y$Y L# xoWanoojҍUq4E_XJŸ\Sͪ )[Ci7etZf;N6dmFFT0 zQWv/|brVGY?;=ܻ +/{[~ekoREJW^|ǥU7D=괲課T~膷?j)hWRf{+ѯ URf{+ѯ UYu:U?P?j{TbZΗâ=E)JܑJRù Iq: + `e[sywWpE{B}63܏չ]Fݵs;} c>{{UB U7l̅ lg˹rŮ7l̅ lg˹rT*?Db{JR- VE#<{}PJuJ_1ozۖ;q/# Hbjq/E; ]wg҆БIKyц&o2za>;Ӷj-mKP)VRʹPbE Ϊ77޵?u5 @q!#b.{*MfNPFJߢk}v䐤\QmHYOIW*iF)W*΋z5ԉ+ylK$ExM: 2%IN-)j:u 4幦I)Z=)槉)Q;$ttu9%ߵ~LD;ZY=V@PJ gYu81Z=6 c9d\][*u2#%PH[mJHSeyk])|wM ,& #BiM%<%%+Jɸk\ʐ 4AZR=r@,RѦuQ]'~È~mL< 7e u̡IBx9#jVW\5Ҷon[vȓl2ظZqJAm # Wk-&_oTeKd)d8K T\j |:7<0&ؖ8{8Yw:68[{nhm86XvKmOJ NOyʽiRCb:Oi$Za+XIua*YJAjY$^ RI &Hj4fS<2)GI=@)JP R3^~mu^Յ3^~mu^3 SK{TkD"fڝ>YrXswf%cs\;^4g3PoaJu_>Y pu_>Y \;aJu_>Y pu_>Y \xoӺ6U3`U3`dž;aJu_>Y pu_>Y \xoӺ6U3`U3`dž;˽˦UTH_hpF]F%tVN3(2E×?EJ:!zR^|%F:ViW6qAڇ!hRE* ZKDѬ]$mͺ,#}2ͩ EqN6J[ [t৴TI s!Ɠ̻CP’ ĵY-v5kZ̥P[@JFХ%* w * E1.&6]ZKL%{ؒJSVnJzBNBs*M_mOG6,)6{e,Ǖ([mvY) a5}+WCKLc~Ɣ5iu(c SaT{ G`W7b>;  [vピz6#qG)bnVqxQ뚚W|w$xg1hϡ[oZ,/XZڻIl42unw(vȟxB1>Sޕ痽?E2"ǥT~4fv"pW3g%H qLgj p^|wM؞θ,MG1Tdd@ImlBR6${uh_5q]߬aNR3 2۞ ⹑V|_mk;KKv#˖야ɵ%[R!kAJ*)VkK[/-nݣ6ZbrJ4)K;;jd׳19mpjឺ g6{{UB U7l̅ lg˹rŮ7l̅ lg˹rT*?Db{JR- VE#<{7w[}W2Ū'Z3smTQ"_´q@ 4Ns)yetl3wR#QFb-Hw^G0A ;Y@%=._C;Ӷj-mKP)VRʹPbE Ϊ77޵?u5 @q!#b.{*MfNPFJߢk}v䐤\QmHYOIW*iF)W*΋z5ԉ+ylK$ExM: 2%INV=4T_tglwp!2gn ArN9`z2q]iiIlƠ7_,+l:fiFAVk]N Gy6CWKPߧke=Eob۔ toV%KF҂Đ!ڮZx5z+>ȇ2r}l:&K{]R6NjO4\.wWE!!l޾)N7#9!8Xh+rӍ\_qGf@mo%IB-[6˵$8:9X3-c8#8!)6KxPvm\2LhPޙ2CQ0ڝyVБ)J<Iʍ/ 9tMhKL E Z؈vRTMFԺVE|j6KHS5$Ka*7)IA*tӚ7Q=ۻa8ӌ[]~jnTU`KilrJ0*=zKSmJ眸#u]y a Lw+aiZdmZnӗ+=PZf\mǙڠoBIRpN*HL\ַE=bS`]9g92:Z_FK q-L%k}ĩv pی~x{er!m[VC\#҂ jWIe)`إ)B)JJRƙ/?6j™/?6j)]YcXb s^v =DzyV;NN]nKac*3_/ PZe)6<=tնmף+<.[]NJ[ܡvGa^IHH%@w s-rlim>6|]((+6㺪ZKe@6 ?&Q~+Z~S8@ s.c^Mj A1̩4L pTv/wgez>7Yѣ6 -TYq Y=!i9ηஏKB KR%cT㄃V ;Kdv ;jm+@iw0]N䯎VQe]pNzݬVGVznrRS댂uaTFw${'fwfݫWG_G iK8]/집?=T-H( 7iҕ$RG{W}> K'ZfqvÁ]YJwyQfE\gujv mHD`T2\;dHӗ $Hcc8ԧpڛ]5JBwwЊ⼇a6YWRv#lN;UU [ݲ͈NGjc y,n R * ;'=µVMG{tx]vU︕W8!|ug%8qrHQW9pp3c ϖ;w[3{^q[ 4]ڶI#J,ASjrn6f*b+w{aNYCe8ۯl*V9IH'q9.M5Gk1ħ>Yqj~T$ VǃPF~]菆]2H-;>P)vRrթR:y 8J`"}M{w;ÑZ80s\e`]7w$6lSa qM$lAu YR1( Z:5~jW[u ѷ#;<Afcx1֠R(B,[4Z!O}f;on#ܐrۘϒ20p317LiKV._m;p6Gd~hn.JW^|XykoLm(,zT[XVᦠؑo2n KPIAE%vnݽ2\m+GC#:NJM,Hܜڵ'#QД+WzBk_fե?2q~ɚK$8:68a[ ]ZiZn)oӖg݃!rb-R Gej_h`s釧Hr[0N-Hxedk -@YH^{*;uC|Q֗_t[/:z\׸-g@!Ć(Lܩ5:]A-+~ ]ۻBVrڏaG{M"?[S< fI>m%\\:.OGR$/Ҟ.OX4*P;ˮ%;0!ZPF9S5~[ ɝ2arJW8ytu9%ߵ~LD;ZY=V@PJ gYu81^,}C~۩nR %[L-JVwB/jiֵw x"]ˑ%XL-uJX߹9n^q/+TY)4 =+^6'U*ܜ s}j-[.]%-͘ CJZA nI9ڜ 'WڹNƮ/v6[Z- Jl=.]9!w Sq‘[ymV66VLdLn4.[d) POy[r}^6!\fZz4p.JՐBP.(jAVF2#7/:UQ:myK-eYqVrS%@yAʛ7d14!Sh)Մ! ԥ% z9ȦycW4Nwc}([hJ{OWI@QZnZb.vٶn,$, iDK+涺6w)<pҖ*KQ."NmeS$Dd41D!B8n&$q(+^4)ږ%Wܖ%(#46.KJ R)f֥,.)JPJRR1~3 ͮڰ~3 ͮچaiz<wvO|4[h^S;[KnSssː'k~Ffw*"pml)Z4g3N4g3Kߧtl)Z4g3N4g3K wF•#F|;#F|;ߧtl)Z4g3N4g3K wF•#F|;#F|;ߧt~w>{tʶikn? $ yrHrGD"OJW^|Jݭ" ԛ3>]P- RTȥDk_׍[펛uUn`e()H*Z֠dLއ2i1mM<˽8 )*IN AKUjj3V% q[q[ m RRyb(}> Ón/\^9d/)`ۭeڔ e ;e :"|Zl9fKeR i=1*@S6G @ jY-ws!,q8e)Z$-!*q%8YNq>)n#}9fw|2Զ(>HaI Evկbkb%OC)LP6*p'ra< ͎]7 ~GNY l۳=;ם5-rX:zQקY~ˬK<-%qH#;3+m?k[CRD$՛oZ÷-q#AqNeRGZ”BpCgzU op}ǩ3=kymǩ3=kymj:QGڨZUt꽇1VgKa}WnH)@aÅȊeRg29rGܮrjwIvn9۝Jqߌ==ʡ*pE{B}63܏չ]FpE{B}63܏չ]FP"\1z=)WpǢd`Y=>_%:L]ԯQjmR8A$r1P r5|8K㮻ҳChHRw $%vq7=0NKfe ̃ a) Oe\qo1ZuUu@Z˚TÈ8i=&'^#EoZ5˾wrHRVZ֮[Q(Vut6ɿGjz,'\4ܣgEֽqicDSֶ%"Q<&s{Jyuy+Pg*f:~3Kayr3T^l NIJ'o=N8G$RscPv }KTK3gԂH Z# A+5#5O~xT'v 'p|=x]yչLiٙCRc>]e* Z܂k+Җ}[Qd4C(R8QQI9#vL5%w(nܒ ^Ky 'Uh[nvۗYu$Ϋ!QuwskNՌ x˽-̃pb"ҥ[noKq!TS+J̅YBz*{i2"ˊnza,Li$$$);{K܍ p(81Czd FjuXBBFT(I',7-Xd8ZSФ%`RT@8 G@ͅ)J)JP R)@cLf]WaLf]W ly|[GWpӹ/ ;U<{dmKuŧl%1•w(l@2rKj6ѕQ.f-nPL#$a$r;99p46Gpmh>[a.xŠIN{qU-%fvNYr Tw(S-M) RZR1r&m Р٘dfTX[&NBP S8Qt *;=\uт[Ɇ\YS U,ҐA []r[WG%MOKR*qA+QUnѥjA;T6Օ4H;.rW+te8 'qʼ"Տ·\7(bdζH̩IqCRB]ٻ=wjwvZnR'eK~;)ObY.]jˋ-W)eϲ?prRV;ޡjи!tx;F yI̵KIڥ6 JcZuMRNMe^#!1҅iT_7Oe!% Q ]N\.m8x>$OSd*r wWKc{nprl 6 ~2a'؜w mӺ7U3`U3`;aJu_>Y pu_>Y \xoӺ?\ =eg.nZv}厡R2<嵫Ruiqv]iZED4m7 //ww"qaD,EnI~ld-aH؂`bQӆHC]5l+ۑҞ[N 1ռRP)i!Jgc-_-DdH9m X8kZe&4LD6ՍRڣO2?4{Y{Tg+/{[~ekoB,zR^|=*-npPlH7{ (i$ ;r7sNnޙ.^|#u葝\cak榖PnNGdZ(hJZ!j5/XjߊrB㸿d`[wPZW`0֭47b731d)QZq/T0Ju4zR^|=)^y{S/{[~ T?Q>ͨd}^=I \gF7mWT^=I \gF7mUds׾ ꎚ=W@f*ժU?P?k:_j+rE)J.DS-%ꔀC<%lVs_K]n)ۙ cr?VwSLv5Sg)yVU1Vݲ,S߷23鱞.~7ݲ,S߷23鱞.~7eR)J|`%[=s#Xߣo\ԝk]ͶYQD|G 4ҁ90m}StΞUJE|V[ܵ!y2@g#Ʒ-FDLdӤ}9-wm[<22 5R/i=rǺ[GZNY^/n[߹9!| S|Dc XیvH_Rk2u4[V>\w$%ej%`WClD?^){;llYbam~xKC,}%L"䫻CnT*DxvuU[vׄ##SWk˷FxѧWQqdv$:gS@+- UsBQ<72#E9'R*)JP?UXS?UC04'\LVZ?Ujthti=uq$0qYת~[wTKrwHR5s6_5.9x A$x׍j7/XJviȻ2@ۄra< o=Vҏh>0;J]&[﫪}]W.w9BtսKگEBķ_•"ᐎ"r#5qv_dI]!euޚGq/KW 0HENGm^ɥ9|Wiwk~A9HDhx3nzd6;ېGiY74m;˚@I˅*Iʓ"BI^쇜՝Wnћ-19wu*CH;aH?}^1tW l۳?Dv}1MElf٬vY,Or݋hv2 e!6$HAsyΫBI7[ٓ$;&K2:eJR2I${Nb=3 ;K;amI}Y:~u-1.vV$hf{{UB U7l̅ lg˹rŮ7l̅ lg˹rT*?Db{JR- VE#<{}PJuJ_1ozۖ;q/# HbcS9%vg7=0NKfe ̃ a) Oe\qo1ZuUu@Z˚TÈ8i=&'^#EoZ5˾wrHRVZ֮[Q(Vut6ɿGjz,'\4ܣgEֽqicDSֶ%"Q<&s{Jyuyzsw |hȲZ[yf&ԗP8̥} @G֖(Npj~N1ﺢjéflZIA+P$d>frl:DPZjď&[IDTh2DG Bt+qI7ee^X?vS/gbrJPV[Xדc0 ?!ia%Մe)R=d M6s՗{oF68WTJtC1!\iaEa*ҝJ}eE!ɍ63!jLgK2%CA'1JR)@)JP R)@cLf]WaLf]W /.|#;սτg{zCч R.)JP R)@)JP R)@)JP R)@+q*+O[WmE=I \gF7mWT^=I \gF7mTuGMj3jUGڨZ/ɇE5\zR"s")uJ@W!ʶ+h9گ7l̅ lg˹rɩ& ٻknv})~3<*]n)ۙ cr?Vw]n)ۙ cr?Vw@Ttp^>[9Gydtl3wR#QFb-Hw^G0A ;YAqƧr?ŪKyцϢ&o2za>;Ӷj-mKP)VRʹPbE Ϊ77޵?u5 @q!#b.{*MfNPFJߢk}v䐤\QmHYOIW*iF)W*΋z5ԉ+ylK$ExM: 2%INM7UZёdɂ:M.6qJ,{S@+-֋B3[ڡʅ)J()JJRR()JJRR()JJRR_vƊR~Vno W3_/ W3_dyu:U?P?j{TbZΗâ=E)JܑJRù Iq: + `e[sywWpE{B}63܏չ]Fݵs;} c>{{UB U7l̅ lg˹rŮ7l̅ lg˹rT*?Db{JR- VE#<{}PJuJ_1ozۖ;q/# HbcS9%vg7=0NKfe ̃ a) Oe\qo1ZuUu@Z˚TÈ8i=&'^#EoZ5˾wrHRVZ֮[Q(Vut6ɿGjz,'\4ܣgEֽqicDSֶ%"Q<&s{Jyuyzsw |hȲZ[yf&ԗP8̥} @G֖(Npj5.Ӽ57pzm7 #٬ֺ~#nM P-N-!qja.iĠ j22JJXZyj}0bTțj\xGgKd %KNTp]t\ņ}#Y[4ĜҭvR\-JPJmx#$(vkIzEԈo-E}Ќ-Ƹ J[ܖ-]JJ5uҦ)5NP5 Ysn2ʥFSe]ƒ s lPWG7\nhpOx:tu-Oc6v%n$) J6wDEŞ<)܍!;ZJT2 ˥(@)@)JP R3^~mu^Յ3^~mu^3 RоoVZ>FT)JP)@)JP R)@)JP R)@)JP RƊ=n4W]sx&g/MrO?͵]Qx&g/MrO?͵S#ϧ|4zaUUWMj3jt&q)JVR8\ZK!)\xK*ح3˿jݲ,S߷23鱞.~7&t'fkSߌڬbwdXnd(gc=][ߴn-wdXnd(gc=][ߴnʥQ%*RxnJ z.pF SStΞUJE|V[ܵ!y2@g#Ʒ-Fw>釧Hr[0N-Hxedk -@YH^{*;uC|Q֗_t[/:z\׸-g@!Ć(Lܩ5:]A-+~ ]ۻBVrڏaG{M"?[S< fI>m%\\:.OGR$/Ҟ.OX4*P;ˮ%;0"݇mi3֭Nħ[4,h[*S@+-νgTt꽇1VU]4zaUYt_cUǨ)[)JPw0r"i.8Tr, bx.ZvȱO~PϦ{+hܚ`NvogҜw9O~3jAbM9w#nW~ѸbM9w#nW~ѻ*GH ^xJU*1=Ù$zǖO}OIN:{Wu+5f-Z9o[rԇxN%s@TLj#ZnL"f'#lý;l!ِa4e!{I쫐= F+Z^T]}nꮨs}SYs^ q ؠR6-2rduh+P|&wnI JYZj=J7~mOP%$rbUr׮-=Lc]HJzĹ=bDX 'Ӯop@#.BT8jkvoy٤ϲ[Z::nDLoccd[wnN8G$RscQtɌCL%댅!eA*),Wk5;ɲ}!-=髵UO:֙b";)PL&,vë{aoeIK X(4d~{눏B"b2BP + (XRPm3cKhCOSK hQBӑIRH &4d4iqm4-9T$X1Sb)BcD#rxNLR/D# pupC!IQS;H*NMouLmSt/p]ìBoc[Ф)o:䝃zB/TXڶDkFj L d 8o2#`[ a25u:::Ѷ͝%4\%Q21%0%7FT)JP)@)JP R)@)JP R)@)JP RƊ=n4W]sx&g/MrO?͵]Qx&g/MrO?͵S#ϧ|4zaUUWMj3jt&q)JVR8\ZK!)\xK*ح3˿jݲ,S߷23鱞.~7&t'fkSߌڬbwdXnd(gc=][ߴn-wdXnd(gc=][ߴnʥQ%*RxnJ z.pF SStΞUJE|V[ܵ!y2@g#Ʒ-Fw>釧Hr[0N-Hxedk -@YH^{*;uC|Q֗_t[/:z\׸-g@!Ć(Lܩ5:]A-+~ ]ۻBVrڏaG{M"?[S< fI>m%\\:.OGR$/Ҟ.OX4*P;ˮ%;0#Su.ӽ/^`GX[8=zclq6ȟn38j #KOrKe'8?5j'`wQdI[aԳ6{LH$2bZyl5VƱjY3ۣݳ+dDxNCS#[9Gydtl3wR#QFb-Hw^G0A ;YAqƧr?ŪKyцϢ&o2za>;Ӷj-mKP)VRʹPbE Ϊ77޵?u5 @q!#b.{*MfNPFJߢk}v䐤\QmHYOIW*iF)W*΋z5ԉ+ylK$ExM: 2%IN5/^_`m[8z[q9ݜs:dIC.-d(wKž p&J\ D#>{5Smɲkyzg&5enpaɆpQQN'3xP+=%<.#Wu},8-$Ǔ(vԤ+6t )bZX5F w<׮Olg_ #/`H}NSũ =7˯(cKhCOSK hQBӑIRH KJF.q .i2 ׍nʷ'1RFwJ&4d4iqm4-9T$X1^ R()JJRR1~3 ͮڰ~3 ͮچaT>֋B3[ڡʅ)J()JJRR()JJRR()JJRR_vƊR~Vno W3_/ W3_dyu:U?P?j{TbZΗâ=E)JܑJRù Iq: + `e[sywWpE{B}63܏չ]Fݵs;} c>{{UB U7l̅ lg˹rŮ7l̅ lg˹rT*?Db{JR- VE#<{}PJuJ_1ozۖ;q/# HbcS9%vg7=0NKfe ̃ a) Oe\qo1ZuUu@Z˚TÈ8i=&'^#EoZ5˾wrHRVZ֮[Q(Vut6ɿGjz,'\4ܣgEֽqicDSֶ%"Q<&s{Jyuy? W{Kg!smԈU !/g(˻S@+-νgTt꽇1VU]4zaUYt_cUǨ)[)JPyemˊX($g{TG5yiQk "=s!C>rGܮvU(EU7l̅ lg˹rŮ7l̅ lg˹rT*?Db{JR- VE#<{}_zJXt(/9h) q+iIP#q?{9N\Ǫshb(2+7=0NKfe ̃ a) Oe\qo1ZuUu@Z˚TÈ8iy-1bk :ꖓ9sJq!S(10oGe 75΋z5ԉ+ylK$ExM: 2%INJخ .6&ϴ";+BiObzAueRB^#P2"sPR9dJI}KxwgmϽ<"DŽPc)]aݍ`-HZKJzPj91lKS CnO▀Va%½k[=JX5Ʒ+V>1dj;O:Ŷy4$қN1 m(S@RSv9rZT)J)JP R)@)JP?UXS?UC0*} okE}AsޭPaB R)@)JP R)@)JP R)@)JP hʻoE~U}Qo)?+77Rf{+ѯ URf{+ѯ U2<:QGڨZUt꽇1VgKa}WnH)@)JP R)@)JP R)@)JP R)@)JP?UXS?UC0"cB^mDgR?Aq[?_[Mo)QdYT+?_[M*f+;GO_[Mo)K!S4~)\Hy>xr"hkyJYJE+;G[R|TW/-v)\Hy>RChR~n}4JE3Gw駊W/-v?_[M*f+;GO_[Mo)K!S4~)\Hy>xr"hkyJYJEM/HӡK)eҖ!ͫRf{+ѯ URf{+ѯ U"νgYtmsvJ in,@JI@ 3Ig>qK};oIx'bX;ÕE=}Z<_ZâwhNϑ'͜|ϫ\A" hU}j1w^gȓ}Zu|>lծ hU}j~zȴy3 ;ӯ3Ig>:>D6szȴy?=}Z<_Ziי$Vy"O9k=}Z<_Z-jL4|ϫNϑ'͜-jO_y5W֦aGzu|>lէ^gȓ}Z_y5W֧GS0#:>D6sӯ3Ig>p_GSEUfޝy"O9iי$V/EUA"3HNϑ'͜|ϫ\A" hU}jfw^gȓ}Zu|>lծ hU}j~zȴy3 ;ӯ3Ig>:>D6szȴy?=}Z<_Ziי$Vy"O9k=}Z<_Z-jL4|ϫNϑ'͜-jO_y5W֦aGzu|>lէ^gȓ}Z_y5W֧GS0#:>D6sӯ3Ig>p_GSEUfWFRԹ*4ho})m-r&N]h<zȴy?=}Z<_Z$ >r}tGEUA"Թ\C}tG|)y hU}j~zȴy.2 >r}tGEUA"Ըt7'@yi>r_y5W֧GR!|)'@yk=}Z<_Z-jKC}tG|)y hU}j~zȴy.2 >r}tGEUA"Ըt7'@yi>r_y5W֧GR!|)'@yk=}Z<_Z-jKC}tG|)y hU}j~zȴy.2 >r}tGEUA"Ըt7'@yi>r_y5W֧GR!|)'@yk=}Z<_Z-jKC}tG|)y hU}j~zȴy.2 >r}tGEUA"Ը_.t#-7- ZZR:y*3_A"^5ݽ\4-d+I*6? $$If!vh5H5 5#vH#v #v:V  0655R5 / 34p$$If!vh5H5 5#vH#v #v:V 0655R5 / 34$$If!vh5H5 5#vH#v #v:V 0655R5 / 34$$If!vh5H5 5#vH#v #v:V 0655R5 / 34$$If!vh5H5 5#vH#v #v:V 0655R5 / 34$$If!vh5H5 5#vH#v #v:V 0655R5 / 34$$If!vh5H5 5#vH#v #v:V 0655R5 / 34UDdFNf  S BA"MainObjectModel1b\Tk/xͻҕI8T n0Tk/xͻҕIPNG  IHDRb$xPLTE3f333f3333f3ffffff3f̙3f3f333f333333333f33333333f33f3ff3f3f3f3333f33̙33333f3333333f3333f3ffffff3f33ff3f3f3f3fff3ffffffffffff3ffff̙fff3fffffff3ffffff3f333f3333f3ffffff3f̙̙3̙f̙̙̙̙3f3f̙333f3̙333f3fff̙fff3f̙̙3f̙3f̙3f333f3333f3ffffff3f̙3f3f9ųbKGDH cmPPJCmp0712HsPIDATx^}v㺒.oϋx&>Oq#9}$DԬuفo`vдlkMO(Z3_U$H"@B"m‡ lk^%Ab@m?  сYw ~,Qg/cX6,zYe:f'b} q}&iˢY Y>==/[Q:p٩jb ;Kx{b=w0-c}\G!=1|z"l>5 F=1 sL%԰p+ -~ȗt%1{/^sL!6($ 't( ,gH#vNKI\sBbD #0<' _$[ RKÌux+w jsR1NgHEτ_&DL/֢Nu8o!vRb4Ddi,\O,_<E9Ey,'~zp 0@Li? %{II8H/Xe,>:' Ѓs 1jSbz QxNGj΋-nsLC^2}Hſ3J!:R?iD.̈́qP([l}*T\ aFyZpZ@1p _as =irRG\!Ym/Qƻ1?$+d|OapZv`0pc2=9 + +{l|=5%IXc(YF'flu&s;E  _'se$\ڷ!؁ź,ﭩ,MO6Š(L, mP)`ZXV*k/|u3KA,VpVqehz  q`^'l|U$:@Ġ-a)cOkN$X-ð?^kfS_d4 ,fYf` bc49AQvr;bX⎱8}Y4P<~ӓ:β 1~/u%>1.aS?@ bO[+Z)ۉm 1 bSb)Rlr$}L_' 17F W.DN4q'Q̱b.<*z$`J~Yl>)1i[?PeB}-TG(w*IYc $?sJ<]Ĕ/,V J TBD uRkŒ$) d*SҔ9 b~\CSL($z.h.9% vRjR)^:)=)|ľ,h 8`^bN'=LԣrUAQ:H#pd[Lf,SGa5|? `vN 7.UN 68>bY"),0J zyJذ 4TYEtFC FaDN9Al} _rܕ@I$0-MD%ǫЯ<`;u [cph=Y@JF_SAqm$Apv&0ewa8.'P]]03#s5cMDŽ:.nӗ){3lbaV ,0 nl1ge ;A adu<-O4"ہź`,8kvj1gnv bPK)IފHuf͢6?]mAQ5fx[Okh[=Y F/QϢ}ܻO_1]ek>:m &!㓝:6Hj+#P@bAxP4nczik-ԞB %5@4n)1׀Ԥ#c ,f,u )5@8]&ދcR7:dĬcP . =tlS*+_k0s9l5tlj_E#O]Y3J[ئa Xl-tlX2_e'bбMW@Аupm[n5* abTI@X0}OR1e\&f[z'j '2*t)2A1N:@l#l`a/²>Vc 7 idߗB9l 0F>qz-$de:(JXWHs񟰸b;-;U'~yfR1H)P~ԟ}v^  ўҗt;~u [?@OfY= O}b0ktv  bYd!=S9CA: sX]z:@b80ZbnaYsXd^4Ij\4h {tOxKꇂ`Zdx=L&ipg8y1rEIFA JQ2@Ēp(u$37xwR46@b#]q& ],pOtCfa3ؙƒD &Pex -N&S8]?{b=hux1M֗ t)d[gcŒSz1eWBu愍ZtD8c[6#E#|7~zb)# aKG&z2cDHwi%5@M2/,1m)>HrkIQ;-8ДOFUJzIY`1L [촶[ aܻf<#LGR(NKKF4XӜrNlr 6ߵ5%K84s ~z\I8 +XlXt%V:dO" v}M/hrdD hI|4r"_ -ĸ?FRf 2Iy1i|l Q1fx.t{jsF I+DGO4Gt\S\:ub(>sJfOumb7 W#IM(]1RWdNrHøk9}oۣt耗aXl1,4?G",aD;5Dsy'coqde)3nN^Qanpʁz^8kɝ/b1Q|/)gJ#yn[?_;J:^aK̥_xȥ+D^,U~N#ISl6чZR򡜭-B| @~V8޾)f./睋8A[xkuBb72"Pڀ 2ɞ^,AYK u3j 8q7,4÷F2M%Tc/OZ׃ŏSkKIid1VPR/IJ׽9Af9il <[l(n3,f$wy>R#DM"OehТ[$Mn 8eW4iE],w6MĖ>'5.=QDmO']J=A,K TsvmGe/`߷H Tyg+>f@l`5(s;2a!`Lx5BH[ٚD:5[cTlYpQ4lqK֢]ʶ=@lk#L#[G+A:x 8}J'5sa!uqѭeckJe*W' Î(Cb?XZ3JR9c}n+|R7l] #`t.0[guʝ;zeuXy\p5hfݕ (WZd;}. e%I ڿ%[ ztuyL2$'k*a^ bt>gh ~M/Ի0H$<bT$Myz%S2˒)lKewwqlw7m:E íB Qdʊ:"^*ATWDDr=~[th;N\RORN4"IEɍaAI_XK*p3Ş¥ ޢx2ͨV2at?6 ^Dk)Ό8Г$,-IȬ.Ch%K9MSĬ!*OSJ|#&UxPLa%k)NISm"H6@&+b ) L`w2u StʦÞ{gay=bo CyActW4# W _f)>%4(*lqmu <.t?2~g =3e* ͭXAe$!.:e!|5cRDӯ4]*)MB0VSľ躤2ᡀ"0B%dSF M)Te嘢ҔyB$9t\,%`h-n T94ʕPJEYb{ʍy7)К Obd0DCr  iZxeUA61Ŵ-&qGZEaG*Rr+Tx2iyrb?ރ utAkt;/sš*l6eܯ齘늦q3+10Q& i>[}s5fErk.V\F.*ImC+k{0QWD|ST!P=8A'0r~=&JB b;  /.O0XeOJWA` 1ɴKcd0hcfU>4/ >~xɭ\?YҸS@=%V>Uz0D^b[D9441]F\úrO̞c ܗ S OJ׌mmC@lS:,mkQ8Osph VP$ϕm|+0٨:1=Łk> bQFj&$#K1 zvQ_ЯFn6wYKrk^UJxU14v=kؓ_O'tE;S<A cXӡ{+6ױj)=PrhId,bcԜ8|nPQii}i[[jmc{XY<"ޝ1[/!Ze@M/ܲn钩hY$+Ě{2ܝhX+0V%u5h)qb۾ :=RbXRGfڳQ޲"jO"AUbSCbͱ2TCC{6?G15a,m9GB=%GQ)LP/K}gI ȫq(F}@l ok ::;& abKc]Xa.X7OX0aڧ{$3;ID _kG$cXv=KU0[gyQj@6Ҡ[U+<$~b-Fv?BYOϐH߳d:I b:W=?(Q@hP^ 9̓0{[} 1csaF4.ǜ:8MNRّ87A Gߥ+5΍'k%eVH20}å  D,o_=IJJ0 ֦px,mmL?bW4sk_4rznC,[a>siѤ;K19T\ 4 m*MPk{NC(]`kPGkX{PNϜ 2 BHI 1/pY~>\m!=ZC,o$\LV% KԀX `tcv]9% c0OPYbL,seerD]'4ՃS4Vtb1F49}x*9 &Ӿ<5+J&2\2 )bݭ?ں !ɪLY? cjt$N&$ $SO~& piKF 4$D*~"H> \2儯g\r!e: U 1eCO]ËMT@D%AX^dE)T:AIɔA3yESyFLv¬Jannl[0˜b6܆vR회Kpl1*ABwW9'T*+Kё_4){vWCQ0.US8AT97fLזLV4bUF1BL4eJ:ʍ}ɅD+dsEdKD*Oc@ =2.0/$gջOvb%(Tnr(\O4(fUI*XԓgJB3IY$ɔ2^'WeQ1f~ PŔy DCW4.TLHHK 'c1x2'e5>e%buAU|i8(! ( ~K:%p~ _[ΰc-6E}WdIO"h~M9%*!CQn槸(u'x53q&Q&e]B%vܽo X~ qri+p0̰jU=v>Zi)=Mb8PTA@ MBX_;w_<% ,Nj:"JŽ'(Kb;bdCfh 5d8ZnZUX7`jG@0X,aV`|a'_fx-WjX9\>gk¯hؓ":oI,Wɔ*=L֢0󠏧H&GCvhN&<ݮVQH/r߻s ACn /(Hiʧ5F`bhhTJߖdәQְ{=4e߲Awo~'rg5­!>nrw}J gD"*H5o| KY]]]@]RQZCocwK*{߻yJ%9XKtK2Jf]\%-ښp .rV7S1vExIf":_}.R>TR[c`"Dҿi.9 }ObiFJj=i>6ON}ZJƭ .yo1QJ֫0G~<+5`n-UN`o- 6;1B;be߆?a Jt֡ȰuQUɥiǏRLMnnLB(0vS&u;cY- RbAt/n_M(PeGqlKZ-{W;9JyIVB 66 N3m19YAv!eEvag{I#pN-z,_5k3J>%c ᪔;o>dma{B:Ev(N<7N<3'֭y5f4; `/Pj*bcdMJ2!2;y"I0'HcniԄe' %UO &*Zͨ!I<`1f[_cqJA^7؁AI.Ra;~mJu: _ ]JFu=oTau HC8?ܮumYBl(4ft?&2"$rB.4pPV -ԮQ=H9ID[" m,t23J|'y*DunwDMB$!8t3FT]JjT>)"q*,NsvQkJK6;v4!Z H9GH*eGӾB Ȥ"QHN:A+@KJCB.?n-f:NzɭTT% f9S=\|P}*X4y4ACNڣ31cbYT<Č]S:艷$AmV]Cup02%/d FZE)] LSd [snl(WĒB''/R>;IuE]FAnLm pg1udCnt8н/ȇWR5#,TMP) ~lUImBAǨ zPr/#eeV8B_Q})X e @XSڲ΅ڷ8vzFƋl b O)4~m!:um4 ѕUVA5&Wb:l?܅ZM,nrVRyTԚ~46bJ*` D,`jn'}<~RZC ̠ey*IһxbEKKAJ_JǺRiW? L5!V @"!F9 G#3]"@HR_s RIU&/_,VȀSJrKq}+đʒe\Xߊ9^6B`MH\`5:i+<]qo\(oLEF"\7(jG" k@ӏ\zbڝ7Bͅ*d5aCM컌}^:ڕJ{y k C r\QbA1H[L:΀š(­2>%j89L]ܪ4Bg8̱s3+\h5NgE;g6f˧?,pZ UC-YWΖv"ȱ10Gw$t\.Cqp~!| oq=.lZc/ !*Yf}= +ʽO֤PJnl+iwrWdqGui bhdlE@oTh7:ĘM'.b&-ghցc ^Gs;qz:AGm2d lIJ5Q#vjnh ?sƌJfYm[5:-}i3YW7ۓu6?=>=?e=!|{x=ƙY3jjiآGhALg\gM{CMpVbiqzbyR/&Jg TyU ;*J@![MW^ƃ7E72e "x3#ߌ=ȶFcc,=a &RPjbh-UH;0XP vl7nwerD0Mê2V}Y{MLaT4Mf&RPHQST_B9G[F=TOPX6 7M%Qݡ`ZpLdםbA\C:~q#W>; 7z(Mֱ˶3U cdX~.ZIt7> q!b;S'rc)VDQc*kCLWo @j削O:sC;-%PXb t/׫iINu6X=j@뺕Cݸ0fMڈ߱ RnT@[?CƲztk<%k)$05CzcXL+J qWDU8:hBgt$uhaV54RVH0bѭD`܇\˷ޘvڹ-(\ޕ fjIZ UvyS;m4\cRdc3#@E`YGcl^,GYc76xaj1Dž_oo˅#= fox{1]TX#1︡Txׯ7_MH|Ik^ fJPڏ&}ruXv"%G\vY`D\fۼvC&Xc} )dzkိ ҢZ՝=y-JQ],1kVl&E2EVI;Y~kVslE07cϻWEi.,ef܍E bmԄʶ?4 -Z9߀p9RR4%|sM_kX FXjH7G* tn_FbWZ b;eňKqj/"a|E_=7[e M5D8OmѣIW*?C` HrW-}8}Qa0skpMF\yR^t( {1-|7vXc:ǖeTLLiJ?;\%O,Ip l;נ00b|Jǜm#ԉ˳`~'H.f-Sj==TY-(=C;xJSyA:mm~S?`lmpt_1:jlYaO+|Pj~D AːZRx"2u&w$u,ݯԖ̜c=7SŕkY*?k[j+,O@Ҋr_n+]݌՘+' ~8X̃Ej9cUk~=Z`p`òki>![{zD on]2:i0;jS,Iͯ? _77fEl}f1d9gTqo#L3 fl|/QWWci_MHUf%,}dz4-+}obĢ"]M7&rOs㌌@ \r ] !e1N#'WH$Yʀ]hR27G8byvK;':-&EE_^|6Jwp [dJa7ʍ@cX͢Y!^* oXX e} I`p,.% ~ݜM+ؿ5VfIr%%#oPx.k?nJFN",F+v'ub(CU!FeuqE# Ȼfm@c b`ڃ)Rzɮ*6fb\rEJlQ.,s*FB"v\QfXmX$ZॿNgk*o#!ih3ĉJY_0ZG*pp<%yVeO} Mذ۽YŨ% .FZ:H#SW5[X c^GWos'eQvl]uK߫s Xl1R~_?ƇVK> eUDᎹuJzCv|g4ZDp=Q/n`wSx2־3a:W/W_``TЌHJ&&8!CP.Ӌ,&TC\b٬MAnc-\vʪ/cVr'aTr_W FehCL4B" KGvV@ј,x1מ[υ%NZh1pYb4Uj )XȠQ"v@'KP)D4 )ݐœ|A,wbmI)fWZk,|##l؅r6s}mg,ݱGT|ؤrVb%.d{gě BE7`*Bi uBmL7/!*Xe7i$+y-og?:sc*/pSw.8hz~`<ۜ4tCfm8'dsz~ _Hk1KuZm7^d-V{*uX\E_P9YA"`Pu]kȫ\ Xc/kWi+xMb㞺a4bvE;hVdj~Ԕ> [N#% չL$2ݜÆ-fR@1}0SEiV8һb\96 gC y?bt-Eibmn -a |4“ &4}g a^yAڢEbjY̵wWL#(N$ڍ:C#s# +b1|IVW# >1DYGd Fw?wi`9ߍEHb}9-u7{VXXhgf5-7=X&d; Qq?xIoS-r۲-m}<8U&0Yy0J!X٦M1"u8ձÝ#aaMF_Pɔs dY]9ixܜ_ kP°ejoødj+ʗͳ63nzŌmҖE뵘ozFZ #YeX!?Ib2Uޚ9Uf5>W0`pګ'š[X`L[* kFc-Z"㜷qBv8xBb툳'KRuF=MŶ-`¸"4Mb+Z߅u/ eYlxz`"KK Ί p3}Ġ"EUBYN $Lm@uӋ-,y[( Ё*fci M_]]5%yoZi url\SyA**t0{[Ļhm ʼ>#8Jϯ4R,N3:nt om_kbL_@%cj:Ђz~{%n̾t/)sEt 4iGGxM85kP܋w xQWW6V5'b3UYapR/4 .H4M_ͯϯL;6qocF˿Oyi=:ۻ +,$Wi&B MG Bl̅8$,Lauf)&Lmu Ѹl! ;D%XHSL92XV\1umL?y9ŠW yދ`)ؾALY16| '} {, AD`lIQ-&6uwӧa1q,d;NFB2oثvL+UE0mVAb r2u%I[ --MՌyc3?s5 c Xx+(ˉ+̘}J!Y'~R,u8:[OF){3Ԩ',}fԸ}u][H٫&PIG F:Ɇo}AIYWiC[}+=@l,Mtz7\R\mJ\ EA"urAņ"^[U;6w3 Zu\Z\iw(_'^)JA#~߲VKgck3H}%J;ˆWQ~:2/sb{&+^XlywۂXof7Uv}9U :KyHX !~8fsbWW},r އŲ>Y,-йb1q*/F:G F} 5*5>)̛\, >V Ȭ6) ʪ˳x]y_Z-6fatLx:#`UNr{`*Ÿ!v_a9Bxv.ڟ+az2ă fiwޣ_;k!b߯ZM12˲l.$ %OKNTm&ͭvw~ R~ހn}Tv?yD tsr 1"ڒ= jEbje'.tLgx+g B-2:aJ b{+ _rCI V I_+濘f7e8@:;@e~bw vb$lGIENDB`1Dd\od  S @A InferenceModel1bM04&)= fh)0!n!04&)= fhPNG  IHDRG&PLTE3f333f3333f3ffffff3f̙3f3f333f333333333f33333333f33f3ff3f3f3f3333f33̙33333f3333333f3333f3ffffff3f33ff3f3f3f3fff3ffffffffffff3ffff̙fff3fffffff3ffffff3f333f3333f3ffffff3f̙̙3̙f̙̙̙̙3f3f̙333f3̙333f3fff̙fff3f̙̙3f̙3f̙3f333f3333f3ffffff3f̙3f3f9ųbKGDH cmPPJCmp0712Hs,IDATx^}v̲%&&dQJě#  tpCR`Bk/0No-9ufIDII4]?]]U-O5Ԣ 2Y&Ip==:;>ۗEQ~˻0 ϨB 8W ?A_Tû -8ͺkOQMn=DrD5ߔ{@]µXcJwpw[,߻†{}]Ly U{2 b7`AJ|#'`~-==jJяA)bxY~ߜjpAe›7r4=GkMtDu'TSsR 3TGBf/1>rt(hI O<=ˈ5P](UG\'8zD0E%|!Ji4&-z Z5#EYO㗬=ՇU=)F/Bx"*h%)K;8Z^:OQ_>T} БRZTZMķFf=BKN/ MU6OSB)˩Q="Ֆ//"PLiU tZd2")vj#Z-UFJ`U&\GgU>㟰MJU +LQiz<5I MCX?F0(`(4㟮X)WI\ЪQ3i:2';QQ*+ 1Fh `LSHQ'hg* Sp| S|-:/6][Q5Q]gj58;~WcP]wA}wt2CY'Pjiu 6 ņl8:0Mq?Q]rLHTfOQbP\tUܥr 7%k`5y+qA@{΁ -#z1?Y!,W3T&~w) XپsW׎n'm 7ۭ"V0]) |ٸ!DBSF<,C2\j`|lzJfþXSbrD Ճ@u"k6U2;A՚zR__]SoP߸Х6E|t#O*)\%|vip'Tu<>ӈd}Gp3ѭ7UTO4o7>P3GYٹB/U[#zD&v*S< ظX՗DaMc4aʹFkA8B:&TV*EXor}Dț%3B(O#?7EI9z4=VZ%IAiAO#wPXN|Rj?`>ѪbF #UtjJbJL=~Gs>3P"M2u`$O+)=C}hNg؀|$YBP>a@~r"1DP]f#/Bzf4($הj!H7@\03-}z%v ay/ћ3lϴ%[zņLS{*+NئӖ%E.RԆ n#Z,Q~š/b9p'h*{TAukf׍GV* &Yi 5T/MqBu yNFY%VmBkBΞ#~lsuZ%Yqۆ T_9q\OnJJu PDUuPEog TS:uh|'TP]x(P-E5 溒3娦<Zw޷ -漗.݂Tsb7b'TZE㦏\|}STOѰ9Jp`}[lH;%c/hK/ѧo*7i}쬫ԽɁASTA5?"R]LܬSgТۜ9_rI>&Xo2~0ÜlwDnciyh,%ƺKnXm /#v#ٱ0sze\:thH-}ObIiAUFNXm^=ҖO0"@xtYҦ7-[?(75F@;ܾpY^)Ȣ;e[[u`saɦg.Zkc埻C\\G-[e&PƁ=ps±CV=q"DU[J\OQRM5 {; voQM6:ŃS*UFM_{/OYӦܱ> ʄm(xHSt{ES?͍&Ot43~@q~Y|YQF<*LOE?P8[}`TquwL7B sWw5Ii5q8BBwjX~4ZswscR\պRUxc>'ϭl2Z%T1&>7wRN]Ph1Gwt^TApL10GpRԞ6wG@<_宀Mc\r,xDy:b$l9-ܢ7r|[*R퐟~[$>ǙmmG Tmʾ!vZ P=8jӖ [)ve!{V[GIc0z1Lr*wǻP+ځ[ު)ƗDž0,EӌR PM '-f=޵6PlROQz4ZzR ºo8A'of~EZV&5X=EPf ` DjM~*΃o,|gOa]O"iz1-ldxD-M]SATF!t`()s!Rv @PN apXk{#$y̡;c/gBvb\G5Q]ѓ&7iȸB^h[N/?你*QS![y5?uL7ўB+6PEV)-l*PU:$9PТr(CrQ@ ( T_W8.r\d Q#PZFa+T\YƤTZqۣ4;iQd103i )OELU9d#~}Xb8M6~0' .?.8|bL#YU57@d|f1$HA=VAoP}x&y4Ƈ_RgSl> W*WՕiۚBHq(3G&+c| &}<C^[Ic=V Lj29ĖhNwAʒT?;jqJ/#Aψ m1/3Δ&HאuˍzyZFZc1`0%LOi޻3ϻz g}M[bz7z&d4dQ#/$|Ɨ0<!apM@1t 9D`T%I*%ifNOcrfC$ˇs""y͵<9|)50Y∉lx.K+D&Yә#rYѵ\Fj]!?0mm;\7*VF.ҵZKavՅ4*L)iTx39(ƦU^ۂskW)@uKm8{X崩:`4#x uADS6p-m _m?ՍG0V+EEfBRxA}%YX4׺%5!W uA̯CǀFSoqqy&? s_dGb/^݇- q|÷4J!Qf8q7<ڵ(r%NZF \&>DF )kdNmCi`xo1' %Y!z5-//;rs!&9mE4sXtjZӟ*J9LIVwP&8p7(M$̖b$#n*Q۲+бH@ ÀBV{bJ5*)\ggmbD)" `BFڢY0RhISa1qɢÏF֮N5 o@sVV$=w׺4#ұ1?`e߁ J'HZG3ގj.j/0H{n-~~iDԡحzǯFPeƐZ)뭤#t>Iz|ҺPtdIVTWH̲h1܏k%-eSz~F{ vvo+]2/3uM+I ٔ焾ي 6piJuHڬoh#IwkចJ±K$ 5#pOQ<8li|㩸a]qQoJ58Ow {Q zyKqk^^¥bc%*5Zvgʁ+n{S p]L&4}}V>c~$C@ܞX\r7?v}/? *x8/+dW+ oum>z&M,e=o.eOb+<0kr+5;?*=kk{ZGgUԁU8˖(uz8֦NꓖT}jS3$n7F?UIsoJU7VV<[! k .j70MG]L`j ^ݥe\0@HrmffS} Hu<*U5*uTw6GM#Ja1'~ft8(bhr`o#*$\ '-dBi3a k6}alQODU46D,HN¦U+L||d̾BP`FU]h"~אq'<>p(^o[ 頠F7v@pyDbǂy݄[ψ.$57va9jށ2uQ35 l6m^OcREJ Hʣp|ȰѰ]KK RVo}oh.ՠ]МPCj#< sNAKx5)/T* `fF@Vlم*•]X50dWp|2=pSwWLY! \+[cȔ30Ȭ-kHM^hGSPp2P 4OjL{IZy"|VA@$L1FEs`C$lf )YUjg5/F©~kȌv\\^fy|1eg8K4TM`̭X(=ӓTCz% C# KtT*X&tt^Hݾ+RVeҷ]QIS}V墳!x lJ3[U@hjQǷqA'a^3^T?;r`F $%#=uǠ>@A+xDUO|ᝩ+-f,FI4RZu$SϚNUP^yB]ok,ڒ*m:_3w,Pds./ZwpNyxI|GQO4.h8ۨʸC{V=([?;܎ϙ& M!Wg6E!>VKy>TuZjFz:I:IObECvb^UiՔNgQpXXWt`lVEeZ cqE9_Nid?P/$F^&iUU1PtBZ~6'yӻ5&^f?C{HUK+xHWh5HP)OjHiUm\L-͙I@OU2aAiNbKXX : 5;>> .e┽Xȑ7#LXjQSW,- `SW  PDջ2_4Du'|]v0_P|S QJ)[wzMk҃ 9欬I1p]:3So,$}6ZRś"PS6|V5Wu^̿1͋}eüNo\nMQ zm-YpGӭޞQ|ԭ9fI~=7>KX̓/u>c (;mz/6wX! \lݝ|$A(fh#|^VvNhU46a 3lI>=`7_ZbSx7*m+,MP~dZt}2O+;@^^:W}W1. aaba.W A.շ/| x"^rF!aj"W$[镾n̯q+}7#{v<TVSXUhz:x.}lfoʇ:gzhNaՌuɑ:f!H~-ﻕ]Z*^W0: Áąkբ%1P3Q)vQUKͺr} oEseׯPmyc:%Zt2;Qu; #]}7X3ZMbu*Xc%b6 8%Z.u3sU$n2iEkOcת.PM:tkeWʪ̛ڕ T@|Nú {h͟ 'U-k cZOek7.e$BFyW4Kkawʇfm}P7.c jԁW Kg6ۉR gC2ZNhe)48 ;%V>0"j` ŮP-{~q5xfK'8R'#eٌ'WW(%ӌX96fY-in b-4eeփUgqHżpICnlqFԵf*+l_GTI ;iV墫Ϳݪ|gTTpj SǮ>\QmY)+kw@ǎꡱ:_dX5#QvIk6lpĐe⨊h7ZfNzGV#֏e>BTI^ U⬕ NZ 47P~oLnZKb&aPnXζT)Tkkj[UcC"+Gm;Q؁_,~(-%ҹCQmS%r-fe*Km)Gn㯖u\xTMT҃qpxT`Sk*t/﷊TZ4iVgQ*^-\NEZf̶ ߨX]FJ5l+ ۨgsqTkvAΣ\SR@4{ƝqljϚ SG@GX?lo'Tf-]n6OVi(5 jj$mKК|$TiߠW'hibXAG $erڥVC5%W#c%'Ci jfYʇCUP}J/k*ah[6oTa *pm!V&!Q0:!UnH"MXGU D/jˌ:.1KQc;SV䄪mTD-u{ā6ߘj!9&" _ͽzf"Ux_!P=m9z)kcMïlk5Q{(VGA 0m|ϫIVjNq$Yj)A-VPM2p+o?j>\ b*)neFs?,1)^FOTNs^VYF+ tUA<5Y$~e!Z* _aImJ+X]B5LG#L|4y7Reť:\/_9ilH Zʇ3Ea ӂz}! >$ə4Ui0K5(,L?f䪤TCd13ث`rAL#.dJYu W0LCAA-A bP2J6]Q*LƸ"97jq 0?J<Z5K@$f`P"A(?T>+2`,f;ϵ@FhPsK>aq>O֟  Oț&(Yx=]hU+BuiKE`7}hBznN2c&p_R:ov卉,kj'*%+`Ia\oG8Qtf'Ѥ?xK9L4rMnlTua(ͽ]PIK$%.& P] - h 6//O~#\nUT&^k=q~Q/O'G ~>)YTOQ5n+׻ki/VMerΆys:𳑓 1`96ܟ[Rl.&RyʂGC^bd*nwWZ7_0'JҰ'fʺ,͝k”@--f CU -:m CVf0ڢ !Aٙ I)j0uo*9xx$LɾZT9_ZSe}!Ё IY@ߒ6| MaG\2w9n@Lf5qM^¤y}D„ց;G$ځv1xҕo_(O6[y}g}fZsՍQӟ>&.ׄZgЛ ߒ? IҝIe. D_ NՒ.E(f&[~=k]rw=Vr`Pf>=Ral$(Z+'y]&ג~}}f!~LTC;Pu'ZP/7JPu'Ztc,wIENDB` Dd9v  S RA,CatCat.bmp (11794 bytes)b A;Αtں & n A;ΑtںPNG  IHDRؑPLTE8HwtRNS#]bKGDH cmPPJCmp0712HsMIDATx^њ( sӛmB8Kv829C??_]Q*Yɿ(}W>wpOy؋u?W>CgX } AmNUn]!< v8jWakAjwr*z}HQaӛ$zN(%=Ft!}sk:=̵șӧb4]~k2}a%.sek!}NF7Xau19إZo(2rqPJB|<.; }k!M$6!hYG|Pcf]DoCh,x,nҵeMh%_TJ267 ه(d $V҈! YnT9YMR[5$2ƇR?E{@r۲ҀA pͩL])7@jC$W%ӿp!0⩒Á&8хZ8IáIjӭ.yÌXx NniNBT{e ޶Rɩ6R!oӢ׍bg6ZQZW]li~lãȰa&7$KQAIX*8c{u"x)+v[c\"]QJ2/9ЬH-lf! e]dBHn3dlšĬ(r [A F8I儌 ;NȷC&;Nh@ə`GCP0@ Cm60i9-Hbqhd KHwBAD2RbJbV> _2:_^ < 8(t=#:ˣ8GP [DǕdBB3NyF,'ICBZ!:+Pa'NC JB:h?,k(acZًJn r=[0v('!99}֙lJ WlY fRh$ދ3-==ǘ KdM Mjƙۧ7PexdJMlk)X\QHo$i`'{uzO4٢zc2' 2I*iJg !CP*~qEk|!'+'3p=VFun)sJ ^0BU<۪he!!!nCЀ s$J5WǵoMk/~S0j]t wCBSBYHy7eϝBBm9>`Q(?yR~!}j YB~Zɘ9cw}.+J&Sl_)m'l;GN,P%+d}f:B9j6kKUnsv(3ń-U57ji%CYM\GPrxt쐀?>wQwYs:wyqVsOA,d?&7y ^?4N]e.ծd~ʈaGIENDB`$$If!vh5E57555#vE#v7#v#v#v:V 406++++,5 5R55/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406++++,555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4$$If!vh5E57555#vE#v7#v#v#v:V 406++++,5 5R55/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406++++,555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh5E575555/#vE#v7#v#v#v#v/:V 406+,5 5R555L/ 34f4 $$If!vh558555u #v#v8#v#v#vu :V 40 6++++,545555J / 34f4$$If!vh558555=58#v#v8#v#v#v=#v8:V 40 6++++,55555/ 34f4$$If!vh558555=58#v#v8#v#v#v=#v8:V 40 6+,5455555/ 34f4$$If!vh558555=58#v#v8#v#v#v=#v8:V 40 6+,5455555/ 34f4$$If!vh558555=58#v#v8#v#v#v=#v8:V 40 6+,5455555/ 34f4$$If!vh558555=58#v#v8#v#v#v=#v8:V 40 6+,5455555/ 34f4Dd\  C *Aexpbul1a"@@b p~k%㸬dxbr/np~k%㸬dxbPNG  IHDR exsRGBPLTERRRcRBcBkkk119֜J֭s΄BBBƜ{{{{9!1RƵBƌkέZJZ޵9{9s9ֵֽƜJ1έJs)έkΌޥƌc9RΔZZZ޵JνZBƔsƔ2P cmPPJCmp0712>1tRNS@fIDATBP#'i&eSL~Qݸ;/u>v1Iar6;/E(:OuUU?i6Y='~ؔݰXJ ]! _ mIENDB`Dd\  C *Aexpbul2a"@@b-w?0Lcذ^ ,2nw?0Lcذ^PNG  IHDR [A0PLTEkޔRscέޜZskޥcޜRΥ޽ֵKytRNS#] cmPPJCmp0712OmGIDATc [DO 5=G gmⱟu^x`}OY3dIENDB`Dd\  C *Aexpbul3a"@@b^u|[ϩ&*x 4n^u|[ϩ&*xPNG  IHDR [AsRGB PLTEfcݎ cmPPJCmp0712 7tRNS@f*IDATWc` QP0%j J,.gMIENDB`Dd V D 3 $expbul1a"@@Dd V E 3 $expbul2a"@@Dd V F 3 $expbul3a"@@Dd V G 3 $expbul1a"@@Dd V H 3 $expbul2a"@@Dd V I 3 $expbul3a"@@AF@F 13Normal $a$CJ_HmH sH tH V@V 13 Heading 1$$ & F(x@&a$ 5CJKH8@8 Heading 2  & F@&F@!F z Heading 3 & Fx@& 56CJ8@18 Heading 4  & F@&^@^ Heading 5*$ & F(@&^`a$6CJT@T Heading 6 & F<@&6CJOJQJkHL@L Heading 7 & F<@& OJQJkHP@P Heading 8 & F<@&6OJQJkHT @T Heading 9 & F<@&6CJOJQJkHDA@D Default Paragraph FontVi@V  Table Normal :V 44 la (k(No List P&P Footnote ReferenceCJH*OJQJkH>O> Author$a$CJOJQJkHPOP Paper-Title $xa$5CJ$OJQJkHFO"F Affiliations$a$ CJOJQJJ2J  Footnote Textp^`pCJOB Bulletp & Fp>Th^`p4 @R4 Footer  !,Ob, E-Mail<Or WAbstractm & Fx>T@& Tf;@ List Number 3p & F8>Th.^8`POP Captions$H#$+Dp/0$a$5CJ<O< References  & FCJ.)@. Page NumberHC@H Body Text Indenth`hRYR  Document Map-D M OJQJ^JB"B Caption$a$5\^JaJtH TB@T Body Text#/H&#$+D@/0$CJ6U@6 Hyperlink >*B*ph4@4 Header ! !`^@"` W Normal (Web)"$dd[$\$a$CJPJ^JaJmH sH JP@2J Body Text 2#$ a$5\FV@AF FollowedHyperlink >*B* phjORj documentlabel%$dd[$\$a$ CJOJPJQJ^JaJmH sH 0Oa0 emailstyle38:Or: Issue'$ !a$6Q@6 Body Text 3(CJ2L@2 Date )$a$CJDOD Style1*@& CJOJQJ\^JaJV>@V Title+$<@&a$5CJ KHOJQJ\^JaJ B'@B Comment ReferenceCJaJ88  Comment Text-CJ@j@ Comment Subject.5\H@H  Balloon Text/CJOJQJ^JaJ.X@. Emphasis6]e@"HTML PreformattedA1$ 2( Px 4 #\'*.25@9xa$CJOJ QJ ^J mHnHu*W@!* Strong5\bOq2bo^ Abstract Text3$ & Fxa$5CJ]mHnHu|OB| 5)Style Normal (Web) + (Latin) Arial Italic46OJQJ]OQ 4.Style Normal (Web) + (Latin) Arial Italic Char"6CJOJQJ]_HmH sH tH Ob 7WEStyle Normal (Web) + (Latin) Arial Char Char Char Char Char Char Char6Oq 6WJStyle Normal (Web) + (Latin) Arial Char Char Char Char Char Char Char CharCJ_HmH sH tH >O> Style HelveticaOJQJVO1V Style Heading 3 + Helvetica9]VOV Style Heading 1 + Helvetica:\O 8Style Heading 1 + Helvetica Right: -0.04" Before: 6 pt;\VO!V Style Heading 2 + Helvetica<\6O6 W Style ArialOJQJO! W9Style Normal (Web) + (Latin) Arial Pattern: Clear (White)>-DM PJ^JaJ:O: WAcknowledgments?O z4Style Normal (Web) + 9 pt After: Auto Kern at 14 pt@CJKHPJ^JaJ                +Z[iuYl ~ 9K~+|)Ohz 2!r!!0"M"!$ %%%%%%&(&i'(H*+/++}-r.//0T111_253`3333*445e67'78+899::);|;;;1<<5=@=u=}==T>m>>>u???@@VA{AA)B[BBBBEFFGZGpGI2IIIWJJKK]LMVNOPxQQhRRR,SmSSS[TjU^VoVWcWWXY[\n];^_```abccdf;gmggijjGkkl mCmmHnnpEpqs1sesss'tutuuvw:xUxVxxx%y&yyyyy-z.zzz{J~~~}`r+ނ['ImՇKb͍)Aӎ2Oˏ_n!1hIřDt~\y~Οڠnߡ ɤa֥¦`ͨը245=ѫQYұs,Q޺ӻԻ߻ >?Z[opʼ˼\w}[(L /gL-57J$%;<QRgh|}n@ "#9:PQ{N<GeOg(5@Yl} B 3H&={Nv~M "  }  (6G9008 #:008 808 #:008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z00 8 808 #z00 8 808 #z00 8 808 #z00 8 808 #z00 8 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z00 8 808 #z00"8 808 #z00"8 808 #z00#8 808 #z00$8 808 #z00%8 808 #z00&8 808 #z00'8 808 #z00(8 808 #z00)8 808 #z00*8 808 #z00+8 808 #z00,8 808 #z00-8 808 +z00.8 808 #z0008 808 #z0008 808 #z0018 808 +z0028 808 #z0048 808 #z0048 808 #z0058 808 #z0068 808 #z0078 808 #z0088 808 #z0098 808 #z00:8 808 #z00;8 808 #z00<8 808 +z00=8 808 #z00?8 808 #z00?8 808 #z00@8 808 #z00A8 808 #z00B8 808 #z00C8 808 #z00D8 808 #z00E8 808 +z00F8 808 #z00H8 808 #z00H8 808 #z00I8 808 #z00J8 808 +z00K8 808 #z00M8 808 #z00M8 808 #z00N8 808 #z00O8 808 +z00P8 808 #z00R8 808 #z00R8 808 #z00S8 808 #z00T8 808 #z00U8 808 #z00V8 808 #z00W8 808 #z00X8 808 #z00Y8 808 +z00Z8 808 #z00\8 808 #z00\8 808 #z00]8 808 #z00^8 808 #z00_8 808 +z00`8 808 #z00b8 808 +z00b8 808 #z00d8 808 +z00d8 808 #z00f8 808 #z00f8 808 #z00g8 808 #z00h8 808 #z00i8 808 #z00j8 808 #z00k8 808 #z00l8 808 #z00m8 808 #z00n8 808 #z00o8 808 #z00p8 808 #z00q8 808 #z00r8 808 #z00s8 808 #z00t8 808 #z00u8 808 #z00v8 808 #z00w8 808 #z00x8 808 #z00y8 808 #z00z8 808 #z00{8 808 #z00|8 808 #z00}8 808 #z00~8 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z00 8 808 #z00 8 808 +z00 8 808 #z00 8 808 #z00 8 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z00!8 808 #z00!8 808 #z00"8 808 #z00#8 808 #z00$8 808 #z00%8 808 #z00&8 808 #z00'8 808 +z00(8 808 #z00*8 808 #z00*8 808 #z00+8 808 #z00,8 808 #z00-8 808 +z00.8 808 #z0008 808 #z0008 808 #z0018 808 #z0028 808 #z0038 808 #z0048 808 #z0058 808 #z0068 808 #z0078 808 #z0088 808 #z0098 808 #z00:8 808 #z00;8 808 #z00<8 808 #z00=8 808 #z00>8 808 #z00?8 808 #z00@8 808 #z00A8 808 #z00B8 808 #z00C8 808 #z00D8 808 +z00E8 808 #z00G8 808 #z00G8 808 #z00H8 808 #z00I8 808 #z00J8 808 #z00K8 808 #z00L8 808 #z00M8 808 +z00N8 808 #z00P8 808 +z00P8 808 #z00R8 808 #z00R8 808 #z00S8 808 +z00T8 808 #z00V8 808 +z00V8 808 #z00X8 808 +z00X8 808 #z00Z8 808 +z00Z8 808 #z00\8 808 +z00\8 808 #z00^8 808 +z00^8 808 #z00`8 808 +z00`8 808 #z00b8 808 +z00b8 808 #z00d8 808 +z00d8 808 #z00f8 808 +z00f8 808 #z00h8 808 #z00h8 808 +z00i8 808 #z00k8 808 #z00k8 808 #z00l8 808 #z00m8 808 #z00n8 808 #z00o8 808 #z00p8 808 +z00q8 808 #z00s8 808 #z00s8 808 #z00t8 808 #z00u8 808 +z00v8 808 #z00x8 808 #z00x8 808 #z00y8 808 #z00z8 808 #z00{8 808 +z00|8 808 #z00~8 808 #z00~8 808 #z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 +z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 #z008 808 Z[iuYl ~ 9K~+|)Ohz 2!r!!0"M"!$ %%%%%%&(&i'(H*+/++}-r.//0T111_253`3333*445e67'78+899::);|;;;1<<<5=@=u=}==T>m>>>u???@@VA{AA)B[BBBBCCbEEFFGZGpGI2IIIWJJKK]LMVNOPxQQhRRR,SmSSS[TjU^VoVWcWWXY[\n];^_```abccdf;gmggijjGkkl mCmmHnnpEpqs1sessst'tutuuvw:x?xMxUxVxdxlxxxxx%y&y=yMyyyyyyyyy-z.z4z?CFIJRZ[^bfgkopswz{ʼ˼\w}[(L /gL-57J $%(,016;<?CFGLQRUX\]bghknqrw|}n@ "#'*./49:=AEFKPQR{N<GeOg(5@Yl} B 3H&={Nv~M "  }  (6luvw*0*0*0*000000)0)0p000000000+00000000030 000000 00 0 0 0 0 0 0 0 , 0 , 0 0 0 0  00|( 0||000( 0||0h) 0h) 0h) 0h0h0h* 0h* 0h* 0h0h( 0||00"00"00" 00" 00" 00" 00"00"( 90||0&0&0&0&( 90||0+0+0+0+( 90||0/0/ 0/ 0/0/0/0/ 0/ 0/( 90||03 03 03 03 03( 90||07( 90||08 :009 "09 "0:9 "0:9 "09"09 "09 "0;9 "0;9 "0;9 "09"09 "09 "0u=9 "0u=9 "09 "0T>9 "0T>9 "0T>9 "0T>9 "09"09 "09"09 "09 "0VA9 "0VA9 "0VA9 "0VA9 "0VA9 "0 9"09 "0 9"09 "0 909 "0 9 "0F9 "0F9 ;00ZG0ZG 0 ZG 0ZG 0ZG 0ZG0ZG 0ZGZG0K0K0K0K0K0K 0ZGZG0xQ0xQ 0xQ 0xQ 0xQ0xQ0xQ0xQ0xQ0xQ 0ZGZG0^V0^V0^V0^V0^V0^V0^V0^V0^V0^V0^V 0ZGZG0` 0` 0`0` 0` 0` 0`0` 0"0;g60;g60;g 0;g;g60j60j60j( 0jj60l- 0- 060m60m( 0jm60p60p 0;gm60s. 60s. 60s 0;gm( 0ss60t# "0t# "0t60t60t60t"0t "0t "0t 0t 60t 60t 60t 0t 60t 60t 60t 0t 60t 60t 60t 0t 60t 60t 60t 0t 60t 60t 60t 0t 60t 60t 60t 0t ( 0ss60z60z( 0ss60~$ "0~$ "0~$ "0~$ "0~$ "0~ 0;gm60r% "0r% "0r% "0r60r0r ;00 <00060/ 0/ 0/ 0/ 0/ 0/ 0 <06060( 0"01010( 00)10)( 0"02102"02102( 0"0_10_( 0"010( 0"0!10!( 0"01060( 0"010 <060t60t60t60t0 60t0 60t0 60t60t( 0tt"0Ο10Ο10Ο10Ο10Ο( 0tt"0ߡ10ߡ60ߡ10ߡ"0ߡ10ߡ60ߡ10ߡ60ߡ10ߡ60ߡ10ߡ60ߡ10ߡ"0ߡ10ߡ60ߡ10ߡ60ߡ10ߡ60ߡ10ߡ( 0tt"0ұ10ұ60ұ10ұ"0ұ10ұ60ұ10ұ <0"0( 0606060"0"0 "0"0 "0"0 "0 "0 0 0 0 0 0 "0 "0 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 60( 060\60\60\60\60\10\60\( 060606060( 060g60g60g60g60g( 0"0106010( 0607"07"07 "07"07 "07"07 "07 "07 07 07 07 07 07 "07 "07 07 "07 "07 "07 07 607 607 07 "07 "07 "07 07 607 607 07 "07 "07 "07 07 607 607 07 "07 "07 "07 07 607 607 07 "07 "07 "07 07 607 607 07 "07 "07 "07 07 607 607 07 "07 "07 "07 07 607 607 07 "07 "07 "07 07 607 607 07 607607107"07( 060"0"0 "0"0 "0"0 "0 "0 0 0 0 0 0 "0 "0 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 "0 "0 "0 0 60 60 0 60606060106010"0( 0"01060601060 <060<60<( 0<<"0( 0<<"010( 0<<60O( "0O( "0O( "0O( "0O( "0O"0O10O"0O10O60O10O( 0<<60106060106010"0106010"0( 0<<"0106010( 0<<"0106010( 0<<"0&10&( 0<<"0106010 <060606060"0 ;00~ 0~ 0~ 0~ 0~ 0~ @0~ @0~ @0~ 0~ 0~ 0~ 0@0@0@0@0@0@00l}00000000000000000000000000000Z[iuYl ~ 9K~+|)Ohz 2!r!!0"M"!$ %%%%%%&(&i'(H*+/++}-r.//0T111_253`3333*445e67'78+899::);|;;;1<<<5=@=u=}==T>m>>>u???@@VA{AA)B[BBBBCCbEEFFGZGpGI2IIIWJJKK]LMVNOPxQQhRRR,SmSSS[TjU^VoVWcWWXY[\n];^_```abccdf;gmggijjGkkl mCmmHnnpEpqs1sessst'tutuuvw:x?xMxUxVxdxlxxxxx%y&y=yMyyyyyyyyy-z.z4z?CFIJRZ[^bfgkopswz{ʼ˼\w}[(L /gL-57J $%(,016;<?CFGLQRUX\]bghknqrw|}n@ "#'*./49:=AEFKPQR{N<GeOg(5@Yl} B 3H&={Nv~M "  }  (6luvw*0*0*0*0000@0)0000@000000+00000000030 000000 00 0 0 0 0 0 0 0 , 0 , 0 , 0 0 0  0 0( 0000( 00^) 0^) 0^) 0^0^0^* 0^* 0^* 0^0^( 00&!0&!0&! 0&! 0&! 0&! 0&!0&!( 900$0$0$0$( 900*0*0*0*( 900.0. 0. 0.0.0.0. 0. 0.( 9002 02 02 02 02( 900 6( 9006 :008 "08 "098 "098 "08"08 "08 "0:8 "0:8 "0:8 "08"08 "08 "0k<8 "0k<8 "08 "0J=8 "0J=8 "0J=8 "0J=8 "08"08 "08"08 "08 "0L@8 "0L@8 "0L@8 "0L@8 "0L@8 "0 8"08 "0 8"08 "0 808 "0 8 "0E8 "0E8 ;00PF0PF 0 PF 0PF 0PF 0PF0PF 0PFPF0I0I0I0I0I0I 0PFPF0nP0nP 0nP 0nP 0nP0nP0nP0nP0nP0nP 0PFPF0TU0TU0TU0TU0TU0TU0TU0TU0TU0TU0TU 0PFPF0_ 0_ 0_0_ 0_ 0_ 0_0_ 0"01f601f601f 01f1f60i60i60i( 0ii60k- 0- 060l60l( 0il60o60o 01fl60 r. 60 r. 60 r 01fl( 0rr60s# "0s# "0s60s60s60s"0s "0s "0s 0s 60s 60s 60s 0s 60s 60s 60s 0s 60s 60s 60s 0s 60s 60s 60s 0s 60s 60s 60s 0s 60s 60s 60s 0s ( 0rr60y60y( 0rr60}$ "0}$ "0}$ "0}$ "0}$ "0} 01fl60h% "0h% "0h% "0h60h0h ;00 <00 0 60 / 0 / 0 / 0 / 0 / 0 / 0  <060܉60܉( 0܉܉"01010( 0܉܉0 10 ( 0܉܉"0)10)"0)10)( 0܉܉"0V10V( 0܉܉"010( 0܉܉"010( 0܉܉"01060( 0܉܉"010 <060k60k60k60k0 60k0 60k0 60k60k( 0kk"0Ş10Ş10Ş10Ş10Ş( 0kk"0֠10֠60֠10֠"0֠10֠60֠10֠60֠10֠60֠10֠60֠10֠"0֠10֠60֠10֠60֠10֠60֠10֠( 0kk"0ɰ10ɰ60ɰ10ɰ"0ɰ10ɰ60ɰ10ɰ <0"0( 0606060"0"0 "0"0 "0"0 "0 "0 0  0  0  0  0 "0 "0 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 60( 060S60S60S60S60S10S60S( 060606060( 060^60^60^60^60^( 0"0106010( 060."0."0. "0."0. "0."0. "0. "0. 0.  0.  0.  0.  0. "0. "0. 0. "0. "0. "0.  0. 60. 60. 0. "0. "0. "0.  0. 60. 60. 0. "0. "0. "0.  0. 60. 60. 0. "0. "0. "0.  0. 60. 60. 0. "0. "0. "0.  0. 60. 60. 0. "0. "0. "0.  0. 60. 60. 0. "0. "0. "0.  0. 60. 60. 0. "0. "0. "0.  0. 60. 60. 0. 60.60.10."0.( 060"0"0 "0"0 "0"0 "0 "0 0  0  0  0  0 "0 "0 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 "0 "0 "0  0 60 60 0 60606060106010"0( 0"01060601060 <0603603( 033"0( 033"010( 03360F( "0F( "0F( "0F( "0F( "0F"0F10F"0F10F60F10F( 03360106060106010"0106010"0( 033"0106010( 033"0106010( 033"010( 033"0106010 <060606060"0 ;0@0@ ;0@0@ :0@0@0@0@0@0@0@00 0@0@0@0@0 0$}00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0%0  eHK.B6LQ[pv"<ݸ3O"e}8hE h +0*5;DKxYk{U%܁-`š +>IZfoz}$0;FQ\gq|@ "*9AP(d I:::EEF2IqIIIJ&JWJJJSSSW_WaWyׅEGqņ!dkćʇׇ/v؈;@MLWvȊHbz+0^g‘rk5x}֖ "h07\TZobk|ʞ&+8v|i֢ݢV_v<6FEJ<B˸ ںܺKP[ Q\Y^`rfqO-uD M1ns~!&me(L>J}>Br[e7~=HhDO^t(M[jGU Q d     c t    >~ jXXXXXXXCCXCXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXCXXXXXXXXXXXXXXXXXXXXCXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX !!R$#kgJԧjb$~k&.b$e3Ljl@Tb$!dQNܱ“7d2$(,^ ob$Y!I{^$b$@V{0Q%Ab$|6* jObuR$葀Y >nv0 f@ 0(  B S  ?#CreatingaMSBNobjectandLoadingaModelCreatinganEmptyModelRenamingandLoadingaModel SavingaModelLoadingtheSameModelTwice CopyingaModelListingAllModelsRemovingModels(Loading_a_ModelandAccessingItsInferenceEFindingaProbabilityRecommendationsDistributionsasTablesAccessingParameterValuesTableDimensionsParentAssignmentsforeachRowEnumeratingtheRowsDefaultParameters(CreatingCausallyIndependentDistributionsParameterstoProbabilitiesPrerequisiteInformationStartingtheExampleDefiningPropertyTypes#AttachingPropertiestoModelsandNodesAccessingapropertyvalueChangingapropertyvalueTestingforapropertyRemovingaproperty _Hlt521061762 _Hlt521061763_PictureBullets)2_!Οߡұ\g7O&b @@)2_!Οߡұ\g7O&k2|F̟$|ǹ lq <7 D~   w ,!    S ̼0 l  $4sD l  L:$  b ĸ Ѹ  \  t ,yJiuk~P:q(D *8EQ__JJ[##*<<E      !"#$&%')(*+-,.0/1tvW=w,BH(6CO[kkYcc)44DGG  !"#$&%')(*+-,.0/1 ?*urn:schemas-microsoft-com:office:smarttags stockticker80*urn:schemas-microsoft-com:office:smarttagsdate;/*urn:schemas-microsoft-com:office:smarttagsaddress>2*urn:schemas-microsoft-com:office:smarttags PersonName9-*urn:schemas-microsoft-com:office:smarttagsplace8,*urn:schemas-microsoft-com:office:smarttagsCity9+*urn:schemas-microsoft-com:office:smarttagsState>**urn:schemas-microsoft-com:office:smarttags PostalCode:.*urn:schemas-microsoft-com:office:smarttagsStreet=*urn:schemas-microsoft-com:office:smarttags PlaceName=*urn:schemas-microsoft-com:office:smarttags PlaceType 2001287DayMonthYear220/.-,+*222/.-,+*--,----2222222-,-,+-,+---,+w ~ ;B",%&K>R>2D=D*E/EEEJEaa9GwوȊ R_ڍ &Ȏ&Ɛΐ_g+36=@MP]ZbyەkvwZmĠ/8@CKN[^k¡ˡΡޡ!,[fgwy6LNY֥,.9CN~#9;FVąhsɪ)4 &,?JnswˬԬ߬ #).9?DEPҭޭ:@18r}ӯۯݰT^Jͳkvĵϵеڵܵ4<OY۶,4HNqxûʻ1:}*<J,4PX[tu &45=/EFfR\Z0:<FU]az$,;EGR'W`bm(v5Lq)+68Mai ,;Abm  *-5?AK`jlw,46>&.=EU]h"&RZ]vwBH^cHMkq&5E(bh-aj@XZb,<FR]%'0 'H\lw}pz%&,2=>O`kl}sps(OWlxy$89?MXYu~N[]bHUR d f o q {   d t v {     o x aefn SYbklU **@=t===@AAB$M2MTTTUUiUCmSmmmmGn}}+Fblڍ'350 ɤΤ- #ƧX[FNγܵƸʸ޺PU}+qr`3:{46!&7;jsilVLO=1ELRUN[d t (Hgj56WX$mn#NObkl333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333u  9tZ(##!$=%*H**+'7799);|;;;1<<<5=@=u==T>m>>>u??@@@BCCEEeFFFZGpGGrH2IKGK]LLO2PQQ[TT``bcmggjGkGnHn|nness ނ'm1D^hQ}k< p'6$bklbkl Eric Horvitz Eric Horvitz Eric Horvitz Eric Horvitz Eric Horvitz Eric Horvitz Eric Horvitz Eric Horvitz Eric Horvitz Carl Kadie-RUP r}VG 8tXUgpQ0A ~gg=b@r,>xî_ Vq#tVjdLq+*)<"  *zEk^-*"22,H9!30W.2o8!$84*9j[M:[qnJ>X~pC9vdD6"QEQ~OGU>*7L$RJBL ^PAx;WY!a zaab԰,eZiz ^jFr(nnmEjyIHun2F!jot}eoIM^zD"r\iz@$/,}.7~p@.@.@..@...@ ....@ .....@ ......@ .......@ ........^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h^`CJOJQJhHh^`OJ QJ o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h88^8`OJ QJ o(hHoh^`OJ QJ ^J o(hHoh  ^ `OJ QJ o(hHh  ^ `OJQJo(hHhxx^x`OJ QJ ^J o(hHohHH^H`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h^`OJ QJ o(hHoh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(^`.^`CJOJ QJ o(opp^p`.@ @ ^@ `.^`.^`.^`.^`.PP^P`.^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h^`OJQJo(hHh^`OJ QJ o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`OJ QJ o(hHoh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(^`.^`.pp^p`.@ @ ^@ `.^`.^`.^`.^`.PP^P`.h^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`.h^`.hpLp^p`L.h@ @ ^@ `.h^`.hL^`L.h^`.h^`.hPLP^P`L.h^`OJ QJ o(hHoh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHhhh^h`OJQJo(hH@.@..@...@ ....@ .....@ ......@ .......@ ........h^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH00^0`CJOJQJo(^`CJOJ QJ o(o^`CJOJ QJ o(^`CJOJ QJ o(pp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(h^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hHh ^`OJQJo(h ^`OJ QJ o(oh pp^p`OJ QJ o(h @ @ ^@ `OJQJo(h ^`OJ QJ o(oh ^`OJ QJ o(h ^`OJQJo(h ^`OJ QJ o(oh PP^P`OJ QJ o(^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h^`.h ^`hH.h pLp^p`LhH.h @ @ ^@ `hH.h ^`hH.h L^`LhH.h ^`hH.h ^`hH.h PLP^P`LhH.^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(^`.^`.pp^p`.@ @ ^@ `.^`.^`.^`.^`.PP^P`.hh^h`CJOJQJo([]^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(^`.^`.pp^p`.@ @ ^@ `.^`.^`.^`.^`.PP^P`.h^`OJQJo(hHh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH^`CJOJQJo(^`CJOJ QJ o(opp^p`CJOJ QJ o(@ @ ^@ `CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(^`CJOJ QJ o(PP^P`CJOJ QJ o(h^`OJ QJ o(hHoh^`OJ QJ ^J o(hHohpp^p`OJ QJ o(hHh@ @ ^@ `OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJ QJ o(hHh^`OJQJo(hHh^`OJ QJ ^J o(hHohPP^P`OJ QJ o(hH0!jo!a#t,}dIM^zzaabqnJ>H9!3^jrnnm@)<"7~pC ~ *q"22k^-[M:BL,eq2o89vdD\izIHun$8_ *9eo;WUgpQg="QEVG OG*7L0Y,>^P--pe^׎z00@\8                  8bs~f1`Q:Rc"                  dJA,چ8'.Le<߄Ќ*jbp~Xo.Nb                                    -QR$dj^ޚ׈z ȋ!з3                                                                                                    '~QfȒA.:j. \h         tH*I9J?Rtj؉V4W޳ըa`VMJ`f]?CFIJRZ[^bfgkopswz{ʼ˼ $%(,016;<?CFGLQRUX\]bghknqrw|} "#'*./49:=AEFKPQlv_@\\prn-corp1\b113-3312-aNe02:winspoolHP LaserJet 8100 Series PS\\prn-corp1\b113-3312-a$S odXrPRIV0''''` \KhC3TRJPHAAV-FRANKT BAPHAAHP LaserJet 8100 Series PS\\prn-corp1\b113-3312-a$S odXrPRIV0''''` \KhC3TRJPHAAV-FRANKT BAPHAAHP LaserJet 8100 Series PSԫ'`@Unknown Gz Times New Roman5Symbol3& z Arial;" Helvetica5& z!Tahoma71 CourierA"GenevaArialU^ MiriamTimes New RomanIArial Unicode MS?5 z Courier New;Wingdings"hFW&vWFUWF ))!4dW 2qHb{Proceedings Template - WORDEnd User Computing Services Carl Kadie-                           ! " # $ % & ' ( ) * + , Oh+'0  , H T ` lxProceedings Template - WORDrocEnd User Computing Servicesnd  Normal.dotm Carl Kadiem13lMicrosoft Word 10.0@bj9@#P@d7@O)՜.+,D՜.+,H hp|  ACMA Proceedings Template - WORD TitleH PTH\HHHHH _PID_HLINKS_AdHocReviewCycleID_EmailSubject _AuthorEmail_AuthorEmailDisplayName_PreviousAdHocReviewCycleID_ReviewingToolsShownOnceA0Gzi_Ohttp://research.microsoft.com/copyright/accept.asp?path=/~horvitz/DT.HTM&pub=2X#\9http://www-ksl.stanford.edu/KSL_Abstracts/KSL-92-66.htmlM_Y6http://msr-webdev/adapt/msbn3/Events/ModelsRemove.htmGUV:http://msr-webdev/adapt/msbn3/Events/ModelNodesRemove.htm~fS7http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm.(P<http://msr-webdev/adapt/msbn3/Events/NodeParentNodesAdd.htm~fM7http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm~fJ7http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm.(G<http://msr-webdev/adapt/msbn3/Events/NodeParentNodesAdd.htm~fD7http://msr-webdev/adapt/msbn3/Events/ModelNodesAdd.htm A9http://msr-webdev/adapt/msbn3/Properties/ExistingKey.htm[E>9http://msr-webdev/adapt/msbn3/Collections/Properties.htm[;9http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm[E89http://msr-webdev/adapt/msbn3/Collections/Properties.htm[E59http://msr-webdev/adapt/msbn3/Collections/Properties.htm[E29http://msr-webdev/adapt/msbn3/Collections/Properties.htm[/9http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm,9http://msr-webdev/adapt/msbn3/Properties/Description.htm{e)<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm-.&7http://msr-webdev/adapt/msbn3/Objects/PropertyType.htm[#9http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm[E 9http://msr-webdev/adapt/msbn3/Collections/Properties.htm9http://msr-webdev/adapt/msbn3/Properties/Description.htm{e<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htmy"<http://msr-webdev/adapt/msbn3/Methods/Add-PropertyTypes.htm{e<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm{e<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm-.7http://msr-webdev/adapt/msbn3/Objects/PropertyType.htmhj 8http://msr-webdev/adapt/msbn3/Overviews/Collections.htmTM2http://msr-webdev/adapt/msbn3/Properties/Prob.htmTM2http://msr-webdev/adapt/msbn3/Properties/Prob.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm6http://msr-webdev/adapt/msbn3/Overviews/Inference.htmVU1http://msr-webdev/adapt/msbn3/Enums/DISTENUM.htm}97http://msr-webdev/adapt/msbn3/Properties/Type-Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm]9http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htmZA:http://msr-webdev/adapt/msbn3/Properties/UsingDefault.htmZA:http://msr-webdev/adapt/msbn3/Properties/UsingDefault.htm5http://msr-webdev/adapt/msbn3/Properties/Default.htm 2http://msr-webdev/adapt/msbn3/Methods/AddDist.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmwo3http://msr-webdev/adapt/msbn3/Properties/Count.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm>http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm"93http://msr-webdev/adapt/msbn3/Collections/Keys.htm>http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm>http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htm@YBhttp://msr-webdev/adapt/msbn3/Methods/Set-Assignment-Evidence.htm8q7http://msr-webdev/adapt/msbn3/Methods/Add-Evidence.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htm\K1http://msr-webdev/adapt/msbn3/Methods/Belief.htmKY1http://msr-webdev/adapt/msbn3/Objects/Engine.htmbd0http://msr-webdev/adapt/msbn3/Objects/State.htm-1/http://msr-webdev/adapt/msbn3/Objects/Node.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htmKY1http://msr-webdev/adapt/msbn3/Objects/Engine.htmKY1http://msr-webdev/adapt/msbn3/Objects/Engine.htmpn0http://msr-webdev/adapt/msbn3/Objects/Model.htmhj{8http://msr-webdev/adapt/msbn3/Overviews/Collections.htm^Nx1http://msr-webdev/adapt/msbn3/Methods/Remove.htmUVu5http://msr-webdev/adapt/msbn3/Collections/Models.htmr9http://msr-webdev/adapt/msbn3/Properties/Description.htmhjo8http://msr-webdev/adapt/msbn3/Overviews/Collections.htmUVl5http://msr-webdev/adapt/msbn3/Collections/Models.htmUVi5http://msr-webdev/adapt/msbn3/Collections/Models.htmRf5http://msr-webdev/adapt/msbn3/Methods/Copy-Model.htmUVc5http://msr-webdev/adapt/msbn3/Collections/Models.htm>-`8http://msr-webdev/adapt/msbn3/Properties/FileFormat.htmL[]6http://msr-webdev/adapt/msbn3/Properties/FileName.htm2%Z/http://msr-webdev/adapt/msbn3/Methods/Save.htm:*W/http://msr-webdev/adapt/msbn3/Methods/Load.htmL[T6http://msr-webdev/adapt/msbn3/Properties/FileName.htm:*Q/http://msr-webdev/adapt/msbn3/Methods/Load.htm@QN2http://msr-webdev/adapt/msbn3/Properties/Name.htm]K9http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htm]H9http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htmUVE5http://msr-webdev/adapt/msbn3/Collections/Models.htm-.B7http://msr-webdev/adapt/msbn3/Objects/PropertyType.htm[E?9http://msr-webdev/adapt/msbn3/Collections/Properties.htm[E<9http://msr-webdev/adapt/msbn3/Collections/Properties.htm-19/http://msr-webdev/adapt/msbn3/Objects/Node.htm-16/http://msr-webdev/adapt/msbn3/Objects/Node.htm_\39http://msr-webdev/adapt/msbn3/Collections/ModelNodes.htm_\09http://msr-webdev/adapt/msbn3/Collections/ModelNodes.htmpn-0http://msr-webdev/adapt/msbn3/Objects/Model.htmpn*0http://msr-webdev/adapt/msbn3/Objects/Model.htmUV'5http://msr-webdev/adapt/msbn3/Collections/Models.htmUV$5http://msr-webdev/adapt/msbn3/Collections/Models.htm(&!/http://msr-webdev/adapt/msbn3/Objects/msbn.htm$:;http://msr-webdev/adapt/msbn3/Overviews/InferenceModel.htmlU6http://msr-webdev/adapt/msbnx/Assessment_Overview.htmvN3http://msr-webdev/adapt/msbnx/Model_Evaluation.htm& 1http://msr-webdev/adapt/msbnx/Model_Creation.htmm? 8http://research.microsoft.com/~dmax/WinMine/Tooldoc.htmdo+http://research.microsoft.com/adapt/MSBNx/do+http://research.microsoft.com/adapt/MSBNx/do+http://research.microsoft.com/adapt/MSBNx/|INumber for a Tech ReportIDnatasamf@microsoft.comtNatasa Milic-FraylingtV72ata  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$%'()*+,-./0123456789:;<=>?@ABCDEFGHIJKTRoot Entry FVData (91Table%9WordDocument\SummaryInformation(DocumentSummaryInformation8&JCompObjj  FMicrosoft Word Document MSWordDocWord.Document.89qRoot Entry F>VData (91Table%9WordDocument\      !"#$%Y[\]^_`abcdefghijklmnopqrstuvwxyz{|}~SummaryInformation(DocumentSummaryInformation8Z9http://msr-webdev/adapt/msbn3/Collections/Properties.htm[;9http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm[E89http://msr-webdev/adapt/msbn3/Collections/Properties.htm[E59http://msr-webdev/adapt/msbn3/Collections/Properties.htm[E29http://msr-webdev/adapt/msbn3/Collections/Properties.htm[/9http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm,9http://msr-webdev/adapt/msbn3/Properties/Description.htm{e)<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm-.&7http://msr-webdev/adapt/msbn3/Objects/PropertyType.htm[#9http://msr-webdev/adapt/msbn3/Methods/Set-Properties.htm[E 9http://msr-webdev/adapt/msbn3/Collections/Properties.htm9http://msr-webdev/adapt/msbn3/Properties/Description.htm{e<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htmy"<http://msr-webdev/adapt/msbn3/Methods/Add-PropertyTypes.htm{e<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm{e<http://msr-webdev/adapt/msbn3/Collections/PropertyTypes.htm-.7http://msr-webdev/adapt/msbn3/Objects/PropertyType.htmhj 8http://msr-webdev/adapt/msbn3/Overviews/Collections.htmTM2http://msr-webdev/adapt/msbn3/Properties/Prob.htmTM2http://msr-webdev/adapt/msbn3/Properties/Prob.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm6http://msr-webdev/adapt/msbn3/Overviews/Inference.htmVU1http://msr-webdev/adapt/msbn3/Enums/DISTENUM.htm}97http://msr-webdev/adapt/msbn3/Properties/Type-Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm]9http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htmZA:http://msr-webdev/adapt/msbn3/Properties/UsingDefault.htmZA:http://msr-webdev/adapt/msbn3/Properties/UsingDefault.htm5http://msr-webdev/adapt/msbn3/Properties/Default.htm 2http://msr-webdev/adapt/msbn3/Methods/AddDist.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmFJ9http://msr-webdev/adapt/msbn3/Collections/Assignment.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmwo3http://msr-webdev/adapt/msbn3/Properties/Count.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htm0&/http://msr-webdev/adapt/msbn3/Objects/Dist.htmQ_9http://msr-webdev/adapt/msbn3/Collections/KeyObjects.htm>http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm"93http://msr-webdev/adapt/msbn3/Collections/Keys.htm>http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm>http://msr-webdev/adapt/msbn3/Collections/Recommendations.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htm@YBhttp://msr-webdev/adapt/msbn3/Methods/Set-Assignment-Evidence.htm8q7http://msr-webdev/adapt/msbn3/Methods/Add-Evidence.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htm\K1http://msr-webdev/adapt/msbn3/Methods/Belief.htmKY1http://msr-webdev/adapt/msbn3/Objects/Engine.htmbd0http://msr-webdev/adapt/msbn3/Objects/State.htm-1/http://msr-webdev/adapt/msbn3/Objects/Node.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htm:67http://msr-webdev/adapt/msbn3/Collections/Evidence.htmKY1http://msr-webdev/adapt/msbn3/Objects/Engine.htmKY1http://msr-webdev/adapt/msbn3/Objects/Engine.htmpn0http://msr-webdev/adapt/msbn3/Objects/Model.htmhj{8http://msr-webdev/adapt/msbn3/Overviews/Collections.htm^Nx1http://msr-webdev/adapt/msbn3/Methods/Remove.htmUVu5http://msr-webdev/adapt/msbn3/Collections/Models.htmr9http://msr-webdev/adapt/msbn3/Properties/Description.htmhjo8http://msr-webdev/adapt/msbn3/Overviews/Collections.htmUVl5http://msr-webdev/adapt/msbn3/Collections/Models.htmUVi5http://msr-webdev/adapt/msbn3/Collections/Models.htmRf5http://msr-webdev/adapt/msbn3/Methods/Copy-Model.htmUVc5http://msr-webdev/adapt/msbn3/Collections/Models.htm>-`8http://msr-webdev/adapt/msbn3/Properties/FileFormat.htmL[]6http://msr-webdev/adapt/msbn3/Properties/FileName.htm2%Z/http://msr-webdev/adapt/msbn3/Methods/Save.htm:*W/http://msr-webdev/adapt/msbn3/Methods/Load.htmL[T6http://msr-webdev/adapt/msbn3/Properties/FileName.htm:*Q/http://msr-webdev/adapt/msbn3/Methods/Load.htm@QN2http://msr-webdev/adapt/msbn3/Properties/Name.htm]K9http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htm]H9http://msr-webdev/adapt/msbn3/Methods/Add-ModelNodes.htmUVE5http://msr-webdev/adapt/msbn3/Collections/Models.htm-.B7http://msr-webdev/adapt/msbn3/Objects/PropertyType.htm[E?9http://msr-webdev/adapt/msbn3/Collections/Properties.htm[E<9http://msr-webdev/adapt/msbn3/Collections/Properties.htm-19/http://msr-webdev/adapt/msbn3/Objects/Node.htm-16/http://msr-webdev/adapt/msbn3/Objects/Node.htm_\39http://msr-webdev/adapt/msbn3/Collections/ModelNodes.htm_\09http://msr-webdev/adapt/msbn3/Collections/ModelNodes.htmpn-0http://msr-webdev/adapt/msbn3/Objects/Model.htmpn*0http://msr-webdev/adapt/msbn3/Objects/Model.htmUV'5http://msr-webdev/adapt/msbn3/Collections/Models.htmUV$5http://msr-webdev/adapt/msbn3/Collections/Models.htm(&!/http://msr-webdev/adapt/msbn3/Objects/msbn.htm$:;http://msr-webdev/adapt/msbn3/Overviews/InferenceModel.htmlU6http://msr-webdev/adapt/msbnx/Assessment_Overview.htmvN3http://msr-webdev/adapt/msbnx/Model_Evaluation.htm& 1http://msr-webdev/adapt/msbnx/Model_Creation.htmm? 8http://research.microsoft.com/~dmax/WinMine/Tooldoc.htmdo+http://research.microsoft.com/adapt/MSBNx/do+http://research.microsoft.com/adapt/MSBNx/do+http://research.microsoft.com/adapt/MSBNx/ Carl Kadie|IPosting a Tech Reportcarlk@microsoft.com='margin-left:.5in'>