4 Hits in 5.0 sec

Societal Biases in Retrieved Contents: Measurement Framework and Adversarial Mitigation for BERT Rankers [article]

Navid Rekabsaz and Simone Kopeinik and Markus Schedl
2021 pre-print
Societal biases resonate in the retrieved contents of information retrieval (IR) systems, resulting in reinforcing existing stereotypes.  ...  In this work, we first provide a novel framework to measure the fairness in the retrieved text contents of ranking models.  ...  ACKNOWLEDGEMENTS Thanks to Klara Krieg for her help on annotating the queries.  ... 
doi:10.1145/3404835.3462949 arXiv:2104.13640v1 fatcat:vjtij4g2tvgorj34hmmssibezy

Rethinking Search: Making Experts out of Dilettantes [article]

Donald Metzler, Yi Tay, Dara Bahri, Marc Najork
2021 arXiv   pre-print
This paper examines how ideas from classical information retrieval and large pre-trained language models can be synthesized and evolved into systems that truly deliver on the promise of expert advice.  ...  crucially they are incapable of justifying their utterances by referring to supporting documents in the corpus they were trained over.  ...  [57, 99] or content addressing (e.g., differentiable neural computers reflect societal biases in that data [4, 36, 76].  ... 
arXiv:2105.02274v1 fatcat:qdghlnv2nnfhnoo6eafdaxqxzy

Sparsity-aware neural user behavior modeling in online interaction platforms [article]

Aravind Sankar
2022 arXiv   pre-print
Modern online platforms offer users an opportunity to participate in a variety of content-creation, social networking, and shopping activities.  ...  In this dissertation, we develop generalizable neural representation learning frameworks for user behavior modeling designed to address different sparsity challenges across applications.  ...  Beyond popularity bias, it is important to quantify and measure other forms of societal biases in online social media and recommendation applications, to ensure appropriate exposure of the protected attributes  ... 
arXiv:2202.13491v1 fatcat:5lhvre4kpzao5ow7gvxa2qnwhq

Fairness of Exposure for Ranking Systems

Ashudeep Singh
Beyond the theoretical evidence in deriving the frameworks and algorithms, empirical results on simulated and real-world datasets verify the effectiveness of the approach on both individual and group-fairness  ...  In these systems, the items to be ranked are products, job candidates, creative content, or other entities that transfer economic benefit.  ...  ACKNOWLEDGEMENTS First and foremost, I am deeply indebted to my Ph.D. advisor Thorsten Joachims whose constant support, guidance, and thoughtful advice is the reason I am able to make this contribution  ... 
doi:10.7298/t38f-gx71 fatcat:djiyjj6vbzc43lkpdzan5wrepa