Universal Search and Recommendation

Search and recommendation are at the heart of how information workers deal with information overload. Recent advances and trends in science show us that search and recommendation will be transformed in the coming years. One example is deep learning, which been revolutionizing search and recommendation through learned representations or embeddings. These trends raise the possibility of creating a universal search and recommendation system that provides push-button customized search and recommendation for any vertical, any data format, and any type of user need. This system would empower every business, regardless of size, to leverage the power of Microsoft’s AI within their search system and provide professionals across many disciplines the ability to improve their everyday workstreams. In this talk, we highlight trends in three areas that are essential to realizing this vision: representing knowledge with customizable universal embeddings, understanding users’ search needs across information communities and professions, and using hyper-personalization to create search systems beyond “”finding”” and help people discover.

These are the slides for a talk given at Microsoft Research Summit 2021, October 2021. The work it presents is joint work from many authors and references are provided in the slides.