Algorithmic Fairness
Tutorials and talks
Ranking and recommendation
- Disparate effects of recommender systems. Invited talk at 2022 ACM-W Greece Winter School based on PhD work of David Solans and Francesco Fabbri.
- Fairness and transparency in ranking Keynote at 2021 BIAS @ ECIR Workshop, Glasgow F&B in IR, and others. [VIDEO|alt.video]
You can cite this earlier version, which appeared in SIGIR Forum, Vol. 52. No. 2, December 2018, pages 64-71.
Social data biases
Work led by Alexandra Olteanu with Fernando Diaz and Emre Kıcıman, published in 2018-2019.
- Alexandra Olteanu, Carlos Castillo, Fernando Diaz, Emre Kıcıman: "Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries" Frontiers in Big Data, Volume 2, July 2019 [ssrn/previous version].
- Alexandra Olteanu, Emre Kıcıman, Carlos Castillo, Fernando Diaz: A Critical Review of Online Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries. Tutorial at WSDM 2018, WWW 2018, SDM 2018.
- Carlos Castillo: "A Brief Overview of Sources and Manifestations of Bias When Working with Social Data." Summary of the tutorial coordinated by Alexandra Olteanu, for the Russian Summer School on Information Retrieval (RuSSIR). August 2018.
Algorithmic fairness
Work led by Sara Hajian with Francesco Bonchi, published in 2016.
- Sara Hajian, Francesco Bonchi, Carlos Castillo: Algorithmic bias: from discrimination discovery to fairness-aware data mining. Tutorial at KDD 2016. [acm|7-pages outline].
- Slides: Parts I and II: discrimination discovery, Parts III and IV: fairness-aware data mining.
- Video: Part I: Introduction and Context, Part II: Discrimination Discovery, Part III: Fairness-Aware Data Mining and Part IV: Challenges and Directions for Future Research.
- Carlos Castillo: Algorithmic Discrimination Talk at MPI Saarbrücken, Germany, April 2019
- Carlos Castillo: "Algorithmic Discrimination." Talk at BCN Analytics Data and Ethics event, April 2018. [VIDEO]
Link-Based Recommender Systems
Work led by PhD student Francesco Fabbri with Yanhao Wang, Marialuisa Croci, Francesco Bonchi, Ludovico Boratto, Michalis Mathioudakis.
- Francesco Fabbri, Yanhao Wang, Francesco Bonchi, Carlos Castillo and Michael Mathioudakis. Rewiring What-to-Watch-Next Recommendations to Reduce Radicalization Pathways. In Proc. TheWebConf 2022. ACM Press. [doi|arxiv|press] Best paper award.
- Francesco Fabbri, Marialuisa Croci, Francesco Bonchi, Carlos Castillo. Exposure Inequality in People Recommender Systems: The Long-Term Effects. To appear in ICWSM 2022. AAAI Press. [arxiv]
- Francesco Fabbri, Francesco Bonchi, Ludovico Boratto, Carlos Castillo. The Effect of Homophily on Disparate Visibility of Minorities in People Recommender Systems. Accepted for publication at ICWSM, AAAI [request by mail]
Risk Assessment of Criminal Recidivism
Work led by PhD student Marzieh Karimi-Haghighi with Songül Tolan, Marius Miron, Emilia Gomez.
- Marzieh Karimi-Haghighi, Carlos Castillo, Songül Tolan, Kristian Lum: Effect of Conditional Release on Violent and General Recidivism: A Causal Inference Study. Journal of Experimental Criminology. Springer. [doi|springer]
- Marzieh Karimi-Haghighi, Carlos Castillo: Enhancing a recidivism prediction tool with machine learning: effectiveness and algorithmic fairness. ICAIL 2021 (Short papers), pp. 210-214. [acm].
- Marzieh Karimi-Haghighi, Carlos Castillo: Efficiency and Fairness in Recurring Data-Driven Risk Assessments of Violent Recidivism. In Proc. of SIGAPP Symposium on Applied Computing (SAC), pp. 994-1002, ACM Press. [doi|acm]
- Marius Miron, Songül Tolan, Emilia Gomez, Carlos Castillo: Evaluating causes of algorithmic bias in juvenile criminal recidivism. Journal on Artificial Intelligence and Law, Springer. [doi]
- Marius Miron, Songül Tolan, Emilia Gomez, Carlos Castillo: Why Machine Learning May Lead to Unfairness: Evidence from Risk Assessment for Juvenile Justice in Catalonia. In Proc. of International Conference on Artificial Intelligence and Law (ICAIL), Montréal, Canada, pp. 83-92. ACM Press. [slides|acm] Best paper award.
University Admissions and Affirmative Action
Work led by PhD students Marzieh Karimi-Haghighi, with Giorgio Barnabo, Corinna Hertweck, Michalis Mathioudakis, Davinia Hernandez-Leo, and Sergio Celis.
- Marzieh Karimi-Haghighi, Carlos Castillo, Davinia Hernández-Leo: A Causal Inference Study on the Effects of First Year Workload on the Dropout Rate of Undergraduates. In AIED 2022. Springer. [springer]. Nominated for Best Paper Award
- Corinna Hertweck, Carlos Castillo, Michael Mathioudakis. Designing Affirmative Action Policies under Uncertainty. Journal of Learning Analytics vol. 9 issue 2, Society for Learning Analytics Research, 2022. [doi|jla]. Based on the MSc Thesis of Corinna Hertweck at Helsinki University.
- Marzieh Karimi-Haghighi, Carlos Castillo, Davinia Hernandez-Leo, Veronica Moreno Oliver: Predicting Early Dropout: Calibration and Algorithmic Fairness Considerations. ADORE Workshop at LAK'21 [arxiv]
- Michael Mathioudakis, Carlos Castillo, Giorgio Barnabo, Sergio Celis: Affirmative Action Policies for Top-k Candidates Selection, With an Application to the Design of Policies for University Admissions. To appear in the ACM Symposium on Applied Computing (SAC), Brno, Czech Republic, March 2020. [arxiv]
- Corinna Hertweck, Carlos Castillo, Michael Mathioudakis: Towards Data-Driven Affirmative Action Policies under Uncertainty. In Fairness, Accountability, and Transparency in Educational Data Cyberspace (FATED) Workshop. [arxiv]
Diversity in Music Recommendarion
Work led by PhD student Lorenzo Porcaro, with Dougal Shakespeare and Emilia Gomez.
- Lorenzo Porcaro, Carlos Castillo, Emilia Gomez, João Vinagre: Fairness and Diversity in Information Access Systems. To be presented at the European Workshop on Algorithmic Fairness (EWAF). [request by mail]
- Lorenzo Porcaro, Emilia Gómez, Carlos Castillo. Assessing the Impact of Music Recommendation Diversity on Listeners: A Longitudinal Study. ACM Transactions on Recommender Systems, 2024. [doi|acm|arxiv]
- Lorenzo Porcaro, Emilia Gómez, Carlos Castillo: Perceptions of Diversity in Electronic Music: the Impact of Listener, Artist, and Track Characteristics. In Proc. of CSCW 2022, Vol. 6, Issue CSCW1, Article 109, pp. 1-26. ACM Press. [doi|acm|arxiv]
- Lorenzo Porcaro, Emilia Gómez, Carlos Castillo. Diversity in the Music Listening Experience: Insights from Focus Group Interviews. Proc. of ACM SIGIR Conference on Human Information Interaction and Retrieval (CHIIR), pp 272-276, ACM Press. [doi|arxiv]
- Dougal Shakespeare, Lorenzo Porcaro, Emilia Gómez, Carlos Castillo: Exploring Artist Gender Bias in Music Recommendation. 2nd Workshop on the Impact of Recommender Systems., 2020.
Sexism and Gender Biases
- Laura Casanovas i Buliart, Priscila Alvarez-Cueva, Carlos Castillo. From The Beatles to Bad Bunny: Sexism in popular music through an automated text analysis. Poster at IC2S2. Copenhagen, Denmark, 2023. [code]
- Júlia Riera, David Solans, Marzieh Karimi-Haghighi, Carlos Castillo, Caterina Calsamiglia: Gender Disparities in Child Custody Sentencing in Spain: a Data Driven Analysis. In Proc. ICAIL 2023. ACM Press. [doi|code and data]
- Marilena Budan, Carlos Castillo: The Coverage of Sexual Violence in Spanish News Media. ICWSM Workshop on Data for the Wellbeing of the Most Vulnerable. [aaai|doi]
Fair Ranking and Recommendation
Work led by PhD student Meike Zehlike with Tom Sühr, Sara Hajian, Ricardo Baeza-Yates, Francesco Bonchi, and Mohamed Megahed
- Fairness and Transparency in Ranking. Keynote at Data and Bias workshop at CIKM. In SIGIR Forum, Vol. 52. No. 2, December 2018, pages 64-71. [updated slides (2021)|sigir forum]
- Meike Zehlike, Carlos Castillo: Reducing Disparate Exposure in Ranking: A Learning To Rank Approach. In WWW Short papers, Taipei, Taiwan. [arxiv]
- Meike Zehlike, Francesco Bonchi, Carlos Castillo, Sara Hajian, Mohamed Megahed and Ricardo Baeza-Yates: FA*IR: A Fair Top-k Ranking Algorithm. In Proc. of CIKM. Singapore, 2017. ACM Press. [acm|arxiv|Python Library|Java Library|ElasticSearch Plug-in]
- Meike Zehlike, Tom Sühr, Carlos Castillo: A Note on the Significance Adjustment for FA*IR with Two Protected Groups. arXiv, 2020. (Corrects the significance adjustment test.)
- David Solans, Francesco Fabbri, Caterina Calsamiglia, Carlos Castillo, Francesco Bonchi: Comparing equity and effectiveness of different algorithms in an application for the room rental market. In Proc. of AIES 2021, pp. 978-988. AAAI/ACM. [video|doi]
Exclusion and Hate Speech
- Valerio Lorini, Javier Rando, Diego Saez-Trumper, Carlos Castillo: Uneven Coverage of Natural Disasters in Wikipedia: the Case of Floods. ISCRAM, Virginia, USA. [press release|iscram|arxiv].
- Ugur Kursuncu, Manas Gaur, Carlos Castillo, Amanuel Alambo, Krishnaprasad Thirunarayan, Valerie Shalin, Dilshod Achilov, I. Budak Arpinar, Amit Sheth. Modeling Islamist Extremist Communications on Social Media using Contextual Dimensions: Religion, Ideology, and Hate. In Proc. of CSCW 2019, Austin, Texas. [arxiv|doi|slides]
- Alexandra Olteanu, Carlos Castillo, Jeremy Boy and Kush Varshney: The Effect of Extremist Violence on Hateful Speech Online. In ICWSM, Stanford, CA, June 2018 [aaai|arxiv|slides|data and code|forbes].
- Diego Sáez-Trumper, Carlos Castillo and Mounia Lalmas: Social Media News Communities: Gatekeeping, Coverage, and Statement Bias (+ supplementary material). In CIKM 2013 (short paper) [acm|slides|mirror|bib]
Other Topics of Algorithmic Bias
- Arpit Merchant, Carlos Castillo: Disparity, Inequality, and Accuracy Tradeoffs in Graph Neural Networks for Node Classification. In Proc. CIKM 2023. ACM Press. [doi|arxiv]
- Alexandra Olteanu, Michael Ekstrand, Carlos Castillo, Jina Suh. Responsible AI Research Needs Impact Statements Too. Pre-print [arxiv|blogpost]
- Teodora Sandra Buda, Joao Guerreiro, Jesus Omana Iglesias, Carlos Castillo, Oliver Smith, Aleksandar Matic: Foundations for Fairness in Digital Health Apps. Frontiers in Digital Health, August 2022. [doi|alt].
- Gemma Galdon Clavell, Mariano Martín Zamorano, Carlos Castillo, Oliver Smith and Aleksandar Matic: Auditing Algorithms: On Lessons Learned and the Risks of Data Minimization. In Proceedings of the AIES 2020 conference. ACM Press. [acm]
Software and data
- Meike Zehlike, Tom Sühr, Carlos Castillo, Ivan Kitanovski: FairSearch: A Tool For Fairness in Ranked Search Results (Pre-print / Demo paper) [arxiv]
- DELTR (Disparate Exposure in Learning To Rank):
- FA*IR (Post-Processing for Fair Ranking):
- https://fairnessmeasures.github.io/
See also: Machnamh - A Machine Learning Toolkit for auditing fairness and facilitating a reflective analysis of ML models.
