{"id":937629,"date":"2023-05-04T09:00:00","date_gmt":"2023-05-04T16:00:00","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?p=937629"},"modified":"2023-05-04T08:17:52","modified_gmt":"2023-05-04T15:17:52","slug":"inferring-rewards-through-interaction","status":"publish","type":"post","link":"https:\/\/www.microsoft.com\/en-us\/research\/blog\/inferring-rewards-through-interaction\/","title":{"rendered":"Inferring rewards through interaction"},"content":{"rendered":"\n<p class=\"has-text-align-center h6\"><em>This research was accepted by the <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/nam06.safelinks.protection.outlook.com\/?url=https%3A%2F%2Ficlr.cc%2FConferences%2F2023&data=05%7C01%7Cv-amelfi%40microsoft.com%7C4f8d3ad02e2e4d7783fe08db4a7e2690%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638185677904637846%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=ZWTF5GUK8hIi3OR0tBRtSucX4Xd309Bl%2BBlqgERCbT0%3D&reserved=0\" target=\"_blank\" rel=\"noreferrer noopener\">2023 International Conference on Learning Representations (ICLR)<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>, which is dedicated to the advancement of the branch of artificial intelligence generally referred to as deep learning.<\/em><\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1400\" height=\"788\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1.jpg\" alt=\"A diagram in which five newspaper icons are lined up in the middle, the first of which is labeled a. An arrow points from the newspaper to an icon of a person above it. The person is labeled x and has a mouse click icon next to it and a thought bubble with the words \u201cI like this!\u201d that\u2019s labeled r. An arrow points from the mouse click icon to a box labeled \u201crecommender system\u201d under the newspapers.\" class=\"wp-image-937641\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1.jpg 1400w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-300x169.jpg 300w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-1024x576.jpg 1024w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-768x432.jpg 768w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-1066x600.jpg 1066w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-655x368.jpg 655w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-343x193.jpg 343w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-240x135.jpg 240w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-640x360.jpg 640w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-960x540.jpg 960w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-1280x720.jpg 1280w\" sizes=\"(max-width: 1400px) 100vw, 1400px\" \/><\/figure>\n\n\n\n<p>Reinforcement learning (RL) hinges on the power of rewards, driving <em>agents<\/em>\u2014or the models doing the learning\u2014to explore and learn valuable actions. The feedback received through rewards shapes their behavior, culminating in effective policies. Yet, crafting reward functions is a complex, laborious task, even for experts. A more appealing option, particularly for the people ultimately using systems that learn from feedback over time, is an agent that can automatically infer a reward function. The <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/interaction-grounded-learning\/\" target=\"_blank\" rel=\"noreferrer noopener\">interaction-grounded learning (IGL) paradigm<\/a> from Microsoft Research enables agents to infer rewards through the very process of interaction, utilizing diverse feedback signals rather than explicit numeric rewards. Despite the absence of a clear reward signal, the feedback relies on a binary latent reward through which the agent masters a policy that maximizes this unseen latent reward using environmental feedback.<\/p>\n\n\n\n<p>In our paper <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/personalized-reward-learning-with-interaction-grounded-learning-igl\/\" target=\"_blank\" rel=\"noreferrer noopener\">\u201cPersonalized Reward Learning with Interaction-Grounded Learning,\u201d<span class=\"sr-only\"> (opens in new tab)<\/span><\/a> which we\u2019re presenting at the <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/iclr.cc\/Conferences\/2023\" target=\"_blank\" rel=\"noreferrer noopener\">2023 International Conference on Learning Representations (ICLR)<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>, we propose a novel approach to solve for the IGL paradigm: <em>IGL-P.<\/em> IGL-P is the first IGL strategy for context-dependent feedback, the first use of inverse kinematics as an IGL objective, and the first IGL strategy for more than two latent states. This approach provides a scalable alternative to current personalized agent learning methods, which can require expensive high-dimensional parameter tuning, handcrafted rewards, and\/or extensive and costly user studies.<\/p>\n\n\n\n<div class=\"wp-block-columns is-layout-flex wp-container-core-columns-is-layout-1 wp-block-columns-is-layout-flex\">\n<div class=\"wp-block-column is-layout-flow wp-block-column-is-layout-flow\">\n<div class=\"annotations \" data-bi-aN=\"citation\">\n\t<ul class=\"annotations__list card depth-16 bg-body p-4 \">\n\t\t<li class=\"annotations__list-item\">\n\t\t\t\t\t\t<span class=\"annotations__type d-block text-uppercase font-weight-semibold text-neutral-300 small\">Publication<\/span>\n\t\t\t<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/interaction-grounded-learning\/\" target=\"_self\" class=\"annotations__link font-weight-semibold text-decoration-none\" data-bi-type=\"annotated-link\" aria-label=\"Interaction-Grounded Learning\" data-bi-aN=\"citation\" data-bi-cN=\"Interaction-Grounded Learning\">\n\t\t\t\tInteraction-Grounded Learning&nbsp;<span class=\"glyph-append glyph-append-chevron-right glyph-append-xsmall\"><\/span>\n\t\t\t<\/a>\n\t\t\t\t\t<\/li>\n\t<\/ul>\n<\/div>\n<\/div>\n\n\n\n<div class=\"wp-block-column is-layout-flow wp-block-column-is-layout-flow\">\n<div class=\"annotations \" data-bi-aN=\"citation\">\n\t<ul class=\"annotations__list card depth-16 bg-body p-4 \">\n\t\t<li class=\"annotations__list-item\">\n\t\t\t\t\t\t<span class=\"annotations__type d-block text-uppercase font-weight-semibold text-neutral-300 small\">Publication<\/span>\n\t\t\t<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/personalized-reward-learning-with-interaction-grounded-learning-igl\/\" target=\"_self\" class=\"annotations__link font-weight-semibold text-decoration-none\" data-bi-type=\"annotated-link\" aria-label=\"Personalized Reward Learning with Interaction-Grounded Learning (IGL)\" data-bi-aN=\"citation\" data-bi-cN=\"Personalized Reward Learning with Interaction-Grounded Learning (IGL)\">\n\t\t\t\tPersonalized Reward Learning with Interaction-Grounded Learning (IGL)&nbsp;<span class=\"glyph-append glyph-append-chevron-right glyph-append-xsmall\"><\/span>\n\t\t\t<\/a>\n\t\t\t\t\t<\/li>\n\t<\/ul>\n<\/div>\n<\/div>\n<\/div>\n\n\n\n<h2 id=\"igl-p-in-the-recommender-system-setting\" class=\"wp-block-heading\">IGL-P in the recommender system setting<\/h2>\n\n\n\n<p>IGL-P is particularly useful for interactive learning applications such as recommender systems. Recommender systems help people navigate increasing volumes of content offerings by providing personalized content suggestions. However, without explicit feedback, recommender systems can\u2019t detect for certain whether a person enjoyed the displayed content. To accommodate, modern recommender systems equate implicit feedback signals with user satisfaction. Despite the popularity of this approach, implicit feedback is not the true reward. Even the click-through rate (CTR) metric, the gold standard for recommender systems, is an imperfect reward, and its optimization naturally promotes clickbait.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"872\" height=\"554\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/recommender-system.png\" alt=\"Interaction-grounded learning (IGL) for the recommender system setting. The recommender system receives features describing a person (x), recommends an item (a), and observes implicit user feedback (y), which is dependent on the latent reward (r) but not r itself, to learn how to better recommend personalized content to the individual.\" class=\"wp-image-937650\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/recommender-system.png 872w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/recommender-system-300x191.png 300w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/recommender-system-768x488.png 768w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/recommender-system-240x152.png 240w\" sizes=\"(max-width: 872px) 100vw, 872px\" \/><figcaption class=\"wp-element-caption\">Interaction-grounded learning (IGL) for the recommender system setting. The recommender system receives features describing a person (x), recommends an item (a), and observes implicit user feedback (y), which is dependent on the latent reward (r) but not <em>r<\/em> itself, to learn how to better recommend personalized content to the individual.<\/figcaption><\/figure>\n\n\n\n<p>This problem has led to the handcrafting of reward functions with various implicit feedback signals in modern recommender systems. Recommendation algorithms will use hand-defined weights for different user interactions, such as replying to or liking content, when deciding how to recommend content to different people. This fixed weighting of implicit feedback signals might not generalize across a wide variety of people, and thus a personalized learning method can improve user experience by recommending content based on user preferences.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n\t<div class=\"border-bottom border-top border-gray-300 mt-5 mb-5 msr-promo text-center text-md-left alignwide\" data-bi-aN=\"promo\" data-bi-id=\"1044939\">\n\t\t\n\n\t\t<p class=\"msr-promo__label text-gray-800 text-center text-uppercase\">\n\t\t<span class=\"px-4 bg-white display-inline-block font-weight-semibold small\">on-demand event<\/span>\n\t<\/p>\n\t\n\t<div class=\"row pt-3 pb-4 align-items-center\">\n\t\t\t\t\t\t<div class=\"msr-promo__media col-12 col-md-5\">\n\t\t\t\t<a class=\"bg-gray-300\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/quarterly-brief\/jun-2024-brief\/\" aria-label=\"Microsoft Research Forum Episode 3\" data-bi-cN=\"Microsoft Research Forum Episode 3\" target=\"_blank\">\n\t\t\t\t\t<img decoding=\"async\" class=\"w-100 display-block\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prodnew\/2024\/06\/RF-Ep3-Recap-BlogHeroFeature-1400x788-1.jpg\" alt=\"Microsoft Research Forum | Episode 3 | panel discussion\" \/>\n\t\t\t\t<\/a>\n\t\t\t<\/div>\n\t\t\t\n\t\t\t<div class=\"msr-promo__content p-3 px-5 col-12 col-md\">\n\n\t\t\t\t\t\t\t\t\t<h2 class=\"h4\">Microsoft Research Forum Episode 3<\/h2>\n\t\t\t\t\n\t\t\t\t\t\t\t\t<p class=\"large\">Dive into the importance of globally inclusive and equitable AI, updates on AutoGen and MatterGen, explore novel new use cases for AI, and more.<\/p>\n\t\t\t\t\n\t\t\t\t\t\t\t\t<div class=\"wp-block-buttons justify-content-center justify-content-md-start\">\n\t\t\t\t\t<div class=\"wp-block-button\">\n\t\t\t\t\t\t<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/quarterly-brief\/jun-2024-brief\/\" class=\"btn btn-brand glyph-append glyph-append-chevron-right\" aria-label=\"Watch on-demand\" data-bi-cN=\"Microsoft Research Forum Episode 3\" target=\"_blank\">\n\t\t\t\t\t\t\tWatch on-demand\t\t\t\t\t\t<\/a>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t\t\t<\/div><!--\/.msr-promo__content-->\n\t<\/div><!--\/.msr-promo__inner-wrap-->\n<span id=\"label-external-link\" class=\"sr-only\" aria-hidden=\"true\">Opens in a new tab<\/span>\t<\/div><!--\/.msr-promo-->\n\t\n\n\n<p>The choice of reward function is further complicated by differences in how people interact with recommender systems. A growing body of work shows that recommender systems don\u2019t provide consistently good recommendations across demographic groups. Previous research suggests that this inconsistency has its roots in user engagement styles. In other words, a reward function that might work well for one type of user might (and often does) perform poorly for another type of user who interacts with the platform differently. For example, <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/csmapnyu.org\/news-views\/news\/who-s-attracted-to-a-clickbait-headline\" target=\"_blank\" rel=\"noreferrer noopener\">older adults have been found to click on clickbait more often<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>. If the CTR is used as an objective, this group of users will receive significantly <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/arxiv.org\/pdf\/2009.09945.pdf\" target=\"_blank\" rel=\"noreferrer noopener\">more clickbait recommendations than the general public<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>, resulting in higher rates of negative user experiences and leading to user distrust in the recommender system.<\/p>\n\n\n\n<p>IGL-P provides a novel approach to optimize content for latent user satisfaction\u2014that is, rewards that a model doesn\u2019t have direct access to\u2014by learning personalized reward functions for different people rather than requiring a fixed, human-designed reward function. IGL-P learns representations of diverse user communication modalities and how these modalities depend on the underlying user satisfaction. It assumes that people may communicate their feedback in different ways but a given person expresses (dis)satisfaction or indifference to all content in the same way. This enables the use of inverse kinematics toward a solution for recovering the latent reward. With additional assumptions that rewards are rare when the agent acts randomly and some negatively labeled interactions are directly accessible to the agent, IGL-P recovers the latent reward function and leverages that to learn a personalized policy.<\/p>\n\n\n\n<h2 id=\"igl-p-successes\" class=\"wp-block-heading\">IGL-P successes<\/h2>\n\n\n\n<p>The success of IGL-P is demonstrated with experiments using simulations, as well as with real-world production traces. IGL-P is evaluated in three different settings:<\/p>\n\n\n\n<ul type=\"1\">\n<li>A simulation using a supervised classification dataset shows that IGL-P can learn to successfully distinguish between different communication modalities.<\/li>\n\n\n\n<li>A simulation for online news recommendation based on <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/data.world\/martinchek\/2012-2016-facebook-posts\" target=\"_blank\" rel=\"noreferrer noopener\">publicly available data from Facebook users<span class=\"sr-only\"> (opens in new tab)<\/span><\/a> shows that IGL-P leverages insights about different communication modalities to learn better policies and achieve consistent performance among diverse user groups (the dataset, created in 2016, consists of public posts from the official Facebook pages of news companies from 2012 to 2016 and aggregated user reactions; because of this aggregation, identifying information can\u2019t be extracted).<\/li>\n\n\n\n<li>A real-world experiment deployed in the Microsoft image recommendation product Windows Spotlight showcases that the proposed method outperforms the hand-engineered reward baseline and succeeds in a practical application serving millions of people.<\/li>\n<\/ul>\n<span id=\"label-external-link\" class=\"sr-only\" aria-hidden=\"true\">Opens in a new tab<\/span>","protected":false},"excerpt":{"rendered":"<p>In reinforcement learning, handcrafting reward functions is difficult and can yield algorithms that don\u2019t generalize well. IGL-P, an interaction-grounded learning strategy, learns personalized rewards for different people in recommender system scenarios.<\/p>\n","protected":false},"author":42183,"featured_media":937641,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"footnotes":""},"categories":[1],"tags":[],"research-area":[13556],"msr-region":[],"msr-event-type":[],"msr-post-option":[243984],"msr-impact-theme":[],"msr-promo-type":[],"msr-podcast-series":[],"msr_event_details":{"start":"","end":"","location":""},"podcast_url":"","podcast_episode":"","msr_research_lab":[199571],"msr_impact_theme":[],"related-publications":[],"related-downloads":[],"related-videos":[],"related-academic-programs":[],"related-groups":[144902],"related-projects":[],"related-events":[],"related-researchers":[{"type":"guest","value":"jessica-maghakian","user_id":"935250","display_name":"Jessica Maghakian","author_link":"<a href=\"https:\/\/www.ams.stonybrook.edu\/~jmaghaki\/index.html\" aria-label=\"Visit the profile page for Jessica Maghakian\">Jessica Maghakian<\/a>","is_active":true,"last_first":"Maghakian, Jessica","people_section":0,"alias":"jessica-maghakian"},{"type":"user_nicename","value":"Cheng Tan","user_id":37953,"display_name":"Cheng Tan","author_link":"<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/people\/chetan\/\" aria-label=\"Visit the profile page for Cheng Tan\">Cheng Tan<\/a>","is_active":false,"last_first":"Tan, Cheng","people_section":0,"alias":"chetan"},{"type":"user_nicename","value":"Paul Mineiro","user_id":33272,"display_name":"Paul Mineiro","author_link":"<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/people\/pmineiro\/\" aria-label=\"Visit the profile page for Paul Mineiro\">Paul Mineiro<\/a>","is_active":false,"last_first":"Mineiro, Paul","people_section":0,"alias":"pmineiro"}],"msr_type":"Post","featured_image_thumbnail":"<img width=\"960\" height=\"540\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-960x540.jpg\" class=\"img-object-cover\" alt=\"A diagram in which five newspaper icons are lined up in the middle, the first of which is labeled a. An arrow points from the newspaper to an icon of a person above it. The person is labeled x and has a mouse click icon next to it and a thought bubble with the words \u201cI like this!\u201d that\u2019s labeled r. An arrow points from the mouse click icon to a box labeled \u201crecommender system\u201d under the newspapers.\" decoding=\"async\" loading=\"lazy\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-960x540.jpg 960w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-300x169.jpg 300w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-1024x576.jpg 1024w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-768x432.jpg 768w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-1066x600.jpg 1066w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-655x368.jpg 655w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-343x193.jpg 343w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-240x135.jpg 240w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-640x360.jpg 640w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1-1280x720.jpg 1280w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/05\/stagant-hero-1400x788-1.jpg 1400w\" sizes=\"(max-width: 960px) 100vw, 960px\" \/>","byline":"<a href=\"https:\/\/www.ams.stonybrook.edu\/~jmaghaki\/index.html\" title=\"Go to researcher profile for Jessica Maghakian\" aria-label=\"Go to researcher profile for Jessica Maghakian\" data-bi-type=\"byline author\" data-bi-cN=\"Jessica Maghakian\">Jessica Maghakian<\/a>, Akanksha Saran, <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/people\/chetan\/\" title=\"Go to researcher profile for Cheng Tan\" aria-label=\"Go to researcher profile for Cheng Tan\" data-bi-type=\"byline author\" data-bi-cN=\"Cheng Tan\">Cheng Tan<\/a>, and <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/people\/pmineiro\/\" title=\"Go to researcher profile for Paul Mineiro\" aria-label=\"Go to researcher profile for Paul Mineiro\" data-bi-type=\"byline author\" data-bi-cN=\"Paul Mineiro\">Paul Mineiro<\/a>","formattedDate":"May 4, 2023","formattedExcerpt":"In reinforcement learning, handcrafting reward functions is difficult and can yield algorithms that don\u2019t generalize well. IGL-P, an interaction-grounded learning strategy, learns personalized rewards for different people in recommender system scenarios.","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/posts\/937629"}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/users\/42183"}],"replies":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/comments?post=937629"}],"version-history":[{"count":24,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/posts\/937629\/revisions"}],"predecessor-version":[{"id":938997,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/posts\/937629\/revisions\/938997"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media\/937641"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=937629"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/categories?post=937629"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/tags?post=937629"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=937629"},{"taxonomy":"msr-region","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-region?post=937629"},{"taxonomy":"msr-event-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-event-type?post=937629"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=937629"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=937629"},{"taxonomy":"msr-promo-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-promo-type?post=937629"},{"taxonomy":"msr-podcast-series","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-podcast-series?post=937629"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}