Computer Science Special Seminar: Konstantinos Stefanidis "Fairness in Data Management"

Fairness in Data Management

Konstantinos Stefanidis
Tampere University
Homepage

Friday, 20 August at 10:30
via Zoom: request the link by email [email protected]
Note! the link will be sent to the CS staff.

Abstract: We increasingly depend on a variety of data-driven algorithmic systems to assist us in many aspects of life. Examples include software systems used in school admissions, housing, pricing of goods and services, credit score estimation, job applicant selection, and sentencing decisions in courts and surveillance. This automation raises concerns about how much we can trust such systems. In this talk, we aim at highlighting our work on defining, modelling, and presenting methods used for ensuring fairness in recommender systems and the problem of entity resolution.  

Specifically, in recommender systems, we address fairness when having access only to information about user-item interactions. We are interested in position and popularity bias and demonstrate the effect of promoting fair results despite of a tolerable decreasing in recommendations quality. Furthermore, we pay special attention on how to ensure fairness in group recommendations and multi-round recommendations. For the entity resolution problem, we focus on controlling the data bias aiming to discover and unify descriptions from different data sources that refer to the same real-world entity. Entity resolution is typically used to improve data quality by reducing data incompleteness (i.e., missing values), redundancy (i.e., duplicate values) or inconsistency (i.e., conflicting values). 

Bio: Kostas Stefanidis is an Associate Professor (Tenured) on Data Science at the Tampere University, Finland. He got his PhD in personalized data management from the University of Ioannina, Greece. His research interests are in the broader area of big data, including personalization and recommender systems, data exploration and large-scale entity resolution and information integration. His publications include more than 100 papers in peer-reviewed conferences and journals, including SIGMOD, ICDE, EDBT, CIKM and ACM TODS, and a book on Entity Resolution in the Web of Data (1979 citations, h-index 25). He has 10 years experience in teaching. He has served in several positions in organizing conferences and workshops, and he regularly serves in the PC of top-tier conferences. He leads the research group on Recommender Systems at Tampere University (https://research.tuni.fi/recsys/).