• Bias

    Bias and Discrimination in Recommender Systems Researches on advancement of recommender systems have lead to the question of fairness. Reseraches have shown that recommender systems can be susceptible to bias. However, the definition of fairness, the source of unfairness, the impact of bias in recommndations, and the mitigation of the situation are still some unresolved issues to work on. In this project, we are focusing on fairness in ranked output by conducting following analysis:

    1. Comparison among the fair ranking metrics
    2. Investigating impact of reviews in introducing bias in recommendations
  • Stereotype

    Stereotype in Information Access Systems

    Information access systems like search engines and recommender systems may perpetuate social stereotypes and reinforce them through their results. In this porject, we aim to address this issue by working on identifying and measuring the stereotypes in retrieved results. Currently we are focusing on the tendency of replicating and manifesting gender stereotypes associated with children’s products through retrived results. We hope this research will contribute towards developing safe web enviroment for children.

  • Review

    Review-based Recommendations

    Reviews are one of the pervasive source of information in generating recommendations. Users express their emotion, feelings towards the items through text reviews. Analyzing and extracting useful information from reviews can be helpful for improving recommendations. For this project we are investigating the usefulness of reviews along with item metadata in mitigating item cold-start problems.