CIIR Talk Series

Title: Gender-related Disparities in Ad Hoc Neural Rankers

Abstract: Recent studies have shown that stereotypical gender biases have found their way into retrieval systems, which can have a potentially harmful impact on the users' judgments when exposed to unfair and biased search results, e.g., problems caused by discrimination against minority groups. This is concerning especially given the fact that not only do a large number of search engine users heavily rely on retrieval systems on a daily basis but also due to the fact that search results often constitute a major component of important practical systems such as recommendation systems, question answering systems, and intelligent assistants, to name a few. In this talk, I will show the extent to which stereotypical gender biases are prevalent in information retrieval gold standard relevance judgment datasets, which then get picked up and acted upon by deep learning-based neural rankers. The talk will also feature our recent works that attempt to de-bias traditional frequency-based retrieval methods as well as neural rankers. In addition, our work on de-biasing gold standard information retrieval datasets will be presented which lead to less biased retrieval outcomes.

I will also use this opportunity to introduce our recent NSERC-funded research program on Responsible Development of Artificial Intelligence and welcome any possible future collaborations.

Bio: Ebrahim Bagheri is a Professor in the Dept of Electrical, Computer and Biomedical Engineering at Ryerson University. He simultaneously holds a Canada Research Chair in Social Information Retrieval and an NSERC Industrial Research Chair in Social Media Analytics. Dr. Bagheri is the recipient of the 2019 NSERC Synergy Award for Innovation in Industry-Academia collaboration and the 2016 Professional Engineers of Ontario Young Engineer of the Year Award. Ebrahim has mentored over 50 graduate students and PDFs from diverse backgrounds, many of whom hold key positions in academia (e.g., Asst. Professors at Windsor, Ryerson and Guelph) and industry (e.g., IBM, Sunlife, RBC, Thomson Reuters). Ebrahim has been the Co-Chair of NSERC Discovery Grant Evaluation Group (1507) in 2021 and 2022.

To attend this talk via Zoom, click here. Participants will need a passcode to attend this event. If you need the passcode for this series, please see the event advertisement on the seminars email list or reach out to Alex Taubman. For any questions about this event with the Center for Intelligent Information Retrieval, please contact Jean Joyce.