Research
Improving how people discover, understand, and use information through advances in search and retrieval.
With the scale of search and recommendation, real-time robust and explainable decision-making is at the heart of search and recommendation systems that work robustly even as the user-base changes, new content appears, and topics rise…
Data trails, recording the way that people interact with content and with each other in an enterprise, are a source of linguistic pragmatics (cues to language meaning implied by social interactions) that can be used…
Search and recommendation is core to many Microsoft offerings—such as Microsoft 365, Microsoft Azure, and Microsoft Bing—and it’s crucial to a growing range of industries, such as biomedicine, retail e-commerce, and legal. Underlying technologies are transforming what we know about search and recommendation and the ability to…
The increasing ability to learn representations of text, code, and even medical chemical compounds is changing what and how people search in every domain. In this panel, we bring together experts from industry and academia…
Extreme classification provides a formulation for large-scale ranking and recommendation problems by treating each item to be ranked or recommended as a separate label in a multi-label classification problem. Scalability and accuracy are well-recognized challenges…
In many domains, user preferences are similar locally within like-minded subgroups of users, but typically differ globally between those subgroups. Local recommendation models were shown to substantially improve top-k recommendation performance in such settings. However,…
Search engines have evolved over time, from initially providing the most relevant URLs to user queries to providing information in response to user queries that summarize content from multiple web documents. Instead of clicking and…
Information overload is a prevalent challenge in many high-value domains. Search in biomedicine, and many other vertical domains, is challenging due to the scarcity of direct supervision from click logs. Self-supervised learning has emerged as…
Human intelligence has the remarkable ability to quickly adapt to new tasks and environments. Starting from a very young age, humans acquire new skills and learn how to solve new tasks by either imitating others’…
Recent research has shown that representations learned from user-user and user-item graphs can be used to improve recommendation performance. In this research, the recommendation model is often trained with representation learning. In project DEEGO, we…
