Advances in bias and fairness in information retrieval : third international workshop, BIAS 2022, Stavanger, Norway, April 10, 2022, revised selected papers / / edited by Ludovico Boratto [and three others]

• Record Nr.: UNISA996478864203316
• Pubbl/distr/stampa: Cham, Switzerland : , : Springer, , [2022]; ©2022
• ISBN: 3-031-09316-X
• Descrizione fisica: 1 online resource (166 pages)
• Collana: Communications in Computer and Information Science ; ; v.1610
• Disciplina: 025.524
• Soggetti: Algorithms; Information retrieval
• Lingua di pubblicazione: Inglese
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Note generali: Description based upon print version of record.
• Nota di bibliografia: Includes bibliographical references and index.
• Nota di contenuto: Intro -- Advances in Bias and Fairness in Information Retrieval: Preface -- Organization -- Contents -- Popularity Bias in Collaborative Filtering-Based Multimedia Recommender Systems -- 1 Introduction -- 2 Related Work -- 3 Method -- 3.1 Defining Popularity -- 3.2 Multimedia Datasets -- 3.3 Recommendation Algorithms and Evaluation Protocol -- 4 Results -- 4.1 RQ1: Relationship Between Item Popularity and Recommendation Frequency -- 4.2 RQ2: Relationship Between Users' Inclination to Popular Items and Recommendation Accuracy -- 5 Conclusion -- References -- The Impact of Recommender System and Users' Behaviour on Choices' Distribution and Quality -- 1 Introduction -- 2 Related Works -- 2.1 Impact of Recommender System on Users' Choice Distribution -- 2.2 Impact of Choice Model on Users' Choice Distribution -- 3 Simulation of Users' Choices -- 4 Experimental Analysis -- 5 Conclusion -- References -- Sequential Nature of Recommender Systems Disrupts the Evaluation Process -- 1 Introduction -- 1.1 Related Works -- 1.2 Notation and Definitions -- 2 Evaluation Systems and Adaptation -- 2.1 Ladder Mechanism -- 2.2 Adversarial Attacks -- 3 Recommender Systems as Sequential Decision Makers -- 3.1 k-NN Recommender System -- 4 Sequence-Aware Adversarial Attacks -- 4.1 Random Window Boosting ; Attack (WBoost) -- 4.2 k-NN Posterior Boosting Attack (PostBoost) -- 5 Experiments -- 5.1 Evaluation on Synthetic Data -- 5.2 Evaluation on ML-100k -- 6 Discussion -- References -- Towards an Approach for Analyzing Dynamic Aspects of Bias and Beyond-Accuracy Measures -- 1 Introduction -- 2 Understanding Long-Term Dynamics -- 3 Case Study -- 4 Conclusion -- References -- A Crowdsourcing Methodology to Measure Algorithmic Bias in Black-Box Systems: A Case Study with COVID-Related Searches -- 1 Introduction -- 2 Preliminaries -- 2.1 Auditing Algorithmic Bias.; 2.2 Auditing Platforms and Search Engines -- 2.3 Crowdsourcing Platform: Amazon Mechanical Turk -- 2.4 Measuring Similarity Among SERPs -- 3 Methodology -- 3.1 Crowdsourcing Search Engine Result Pages -- 3.2 Queries -- 4 Results and Discussion -- 4.1 Collected Data -- 4.2 Demographics -- 4.3 Do Different Participants Get Different Search Results for the Same Queries? -- 5 Do Results Vary Between Positive and Negative Query Formulations? -- 6 Conclusion -- References -- The Unfairness of Active Users and Popularity Bias in Point-of-Interest Recommendation -- 1 Introduction -- 2 Related Work -- 3 Experimental Setup -- 3.1 Datasets -- 3.2 Evaluation Metrics -- 4 Popularity Bias in POI Data (RQ1) -- 4.1 Consumption Distribution of POIs -- 4.2 User Profiles and Popularity Bias -- 5 Results and Discussion -- 5.1 Trade-Off on Accuracy, User and Item Fairness (RQ2) -- 5.2 Popularity Bias in POI Recommendation (RQ3) -- 6 Conclusion and Future Work -- References -- The Unfairness of Popularity Bias in Book Recommendation -- 1 Introduction -- 2 Popularity Bias in Data -- 2.1 Reading Distribution of Books -- 2.2 User Profile Size and Popularity Bias in Book Data -- 3 Popularity Bias in Book Recommendation -- 3.1 Recommendation of Popular Books -- 3.2 Popularity Bias for Different User Groups -- 3.3 Unfairness of Popularity Bias vs. Personalization -- 4 Discussion -- 5 Conclusion and Future Work -- References -- Mitigating Popularity Bias in Recommendation: Potential and Limits of Calibration Approaches -- 1 Introduction and Background -- 2 Research Methodology -- 2.1 Baseline Algorithms and Re-ranking Algorithms -- 2.2 Metrics -- 2.3 Datasets -- 3 Results -- 4 Summary and Future Work -- References -- Analysis of Biases in Calibrated Recommendations -- 1 Introduction -- 2 Related Work -- 2.1 Biases in Recommender Systems -- 2.2 Calibration -- 3 Preliminaries.; 3.1 Data -- 3.2 Recommendation Models -- 3.3 Calibration -- 4 Results -- 4.1 Analysis of Bias in the Calibration Algorithm -- 4.2 Analysis of Recommendation Accuracy in Calibration Process -- 5 Conclusions and Future Work -- References -- Do Perceived Gender Biases in Retrieval Results Affect Relevance Judgements? -- 1 Introduction -- 2 Related Work -- 3 Experiment Setup -- 4 Results and Discussion -- 4.1 Gender-Agnostic Experiments -- 4.2 Gender-Specific Experiments -- 4.3 Discussion -- 4.4 Limitations of the Experiments -- 5 Conclusion and Future Work -- References -- Enhancing Fairness in Classification Tasks with Multiple Variables: A Data- and Model-Agnostic Approach -- 1 Introduction -- 2 Background Knowledge and Related Work -- 2.1 Fairness Definition -- 2.2 Related Works -- 3 Debiaser for Multiple Variables (DEMV) -- 4 Experimental Analysis -- 4.1 Employed Datasets -- 4.2 Experimental Results -- 5 Conclusion and Future Work -- References -- Keyword Recommendation for Fair Search -- 1 Introduction -- 2 Related Work -- 3 Objective -- 4 Proposed Method -- 4.1 Problem Setup -- 4.2 The FairKR Framework and Implementation -- 4.3 Limitations -- 5 Experimental Setup -- 5.1 Data -- 5.2 Search Engine -- 6 Analysis and Discussion -- 7 Conclusion and Future Work -- References -- FARGO: ; A Fair, Context-AwaRe, Group RecOmmender System -- 1 Introduction -- 2 Related Work -- 3 The Proposed Method -- 3.1 CtxInfl -- 3.2 FARGO -- 4 Experimental Results -- 4.1 TV Dataset -- 4.2 Music Dataset -- 5 Conclusions -- References -- Author Index.