Abstract

Search systems aim to answer a wide variety of requests from many different users by locating and presenting the most relevant information from large, usually unstructured, data collections. The nature of such tasks makes searching information a complex problem involving challenges from information retrieval, machine learning, human-computer interaction, and many other fields.

This talk will introduce research for latest challenges in information retrieval from three aspects. First, this talk will cover the limited open training data problem in information retrieval and introduce a new solution called similarity-based distant supervision, which automatically generates large-scale training instances for text retrieval tasks only using publicly accessible resources. Second, the talk will also discuss human factors in information search and examine a fundamental question in particular: what counts as good search systems and results for users and how to improve search systems accordingly. Third, this talk will introduce novel applications of using human behavioral data to refine and evaluate search and conversational systems at different settings. The talk will conclude by envisioning the future trends and challenges in information retrieval and search.

Biography

Jiepu Jiang is a Ph.D. candidate in Computer Science at the University of Massachusetts Amherst. He also has a Ph.D. degree in Library & Information Science from the University of Pittsburgh. His research aims to help people better access and use information, especially from large unstructured text corpora and in interaction and exploratory settings. He has been regularly publishing and serving in top information retrieval and data mining conferences such as SIGIR, WSDM, WWW, and CIKM. He received the best student paper award from CHIIR 2017 for his work on understanding dynamics of search result judgments in information retrieval.