Towards Explainable Information Retrieval Models for Professional Search Tasks
Guest Speaker: Yue Wang, Assistant Professor at the University of North Carolina, School of Information and Library Science
Date/Time: Nov 17 at 3 p.m., Central Time
In professional search tasks such as precision medicine literature search, a query often involves multiple aspects. To assess the relevance of a document, a searcher would painstakingly validate each aspect in the query and follow a task-specific logic to make a relevance decision. In such scenarios, we say the searcher makes a structured relevance judgment, as opposed to the conventional univariate (binary or graded) relevance judgment. Ideally, a search engine can support the searcher’s workflow and follow the same steps to predict document relevance. This approach may not only yield highly effective retrieval models, but also open up opportunities for the model to explain its decision in the same ‘lingo’ as the searcher.
In this talk, Dr. Wang will discuss recent work on explainable retrieval models that emulate how medical experts make structured relevance judgments. Using data from the TREC Precision Medicine literature search track (2017-2019), he found that a simple, explainable, and label-efficient model can consistently perform as well as complex, black-box, and data-hungry learning-to-rank models. These results suggest that leveraging the structure in professional search queries is a promising direction towards building explainable search systems to support professional search tasks.
Speaker: Yue Wang is an assistant professor in the School of Information and Library Science at the University of North Carolina at Chapel Hill. His research interests include text data mining, machine learning, information retrieval, natural language processing, and health informatics. His recent work focuses on designing and evaluating interactive and interpretable machine learning algorithms that can help scientists gain knowledge from large unstructured text. These algorithms have been applied to a wide range of data mining tasks, including high-recall search, computer-assisted content analysis, and clinical natural language processing. He publishes in prestigious venues in computer and information sciences, including SIGIR, WSDM, ACL, KDD, AMIA, and JAMIA. He serves as a regular program committee member for WSDM, SIGIR, and WWW. He received the Best Paper Award in WSDM 2016 and Outstanding Program Committee Member Award in WSDM 2016, 2019, and 2020.
Thank you to iSchool assistant professor Jiepu Jiang for inviting Dr. Wang!