Dagstuhl Perspectives Workshop 19482
Diversity, Fairness, and Data-Driven Personalization in (News) Recommender System
( Nov 24 – Nov 29, 2019 )
- Abraham Bernstein (Universität Zürich, CH)
- Claes De Vreese (University of Amsterdam, NL)
- Natali Helberger (University of Amsterdam, NL)
- Wolfgang Schulz (Universität Hamburg, DE)
- Katharina A. Zweig (TU Kaiserslautern, DE)
- Shida Kunz (for scientific matters)
- Annette Beyer (for administrative matters)
- Diversity, Fairness, and Data-Driven Personalization in (News) Recommender System (Dagstuhl Perspectives Workshop 19482). Abraham Bernstein, Claes De Vreese, Natali Helberger, Wolfgang Schulz, and Katharina A. Zweig. In Dagstuhl Reports, Volume 9, Issue 11, pp. 117-124, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2020)
- Diversity in News Recommendation (Dagstuhl Perspectives Workshop 19482). Abraham Bernstein, Claes de Vreese, Natali Helberger, Wolfgang Schulz, Katharina Zweig, Christian Baden, Michael A. Beam, Marc P. Hauer, Lucien Heitz, Pascal Jürgens, Christian Katzenbach, Benjamin Kille, Beate Klimkiewicz, Wiebke Loosen, Judith Moeller, Goran Radanovic, Guy Shani, Nava Tintarev, Suzanne Tolmeijer, Wouter van Atteveldt, Sanne Vrijenhoek, and Theresa Zueger. In Dagstuhl Manifestos, Volume 9, Issue 1, pp. 43-61, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2021)
As people increasingly rely on online media and recommender systems to consume information, engage in debates, and form their political opinions, the design goals of online media and news recommenders have wide implications for the political and social processes that take place online and offline. Current recommender systems have been observed to promote personalization and more effective forms of informing, but also to narrow the user’s exposure to diverse content. Concerns about echo-chambers and filter bubbles highlight the importance of design metrics that can successfully strike a balance between fairness and accurate recommendations that respond to individual information needs and preferences. This balance is additionally complicated by concerns about missing out important information, context and the broader cultural and political diversity in the news.
The goal of this Dagstuhl Perspectives Workshop is to develop a broader, more sophisticated vision of the future of personalized recommenders – a vision that should be developed as the result of a collaborative effort by different areas of academic research (media studies, computer science, law and legal philosophy, communication science, political philosophy, and democratic theory). Collaborating across disciplines, the workshop aims at developing a new perspective on diversity and fairness in recommender systems to address the societal and technical challenges current systems raise. This perspective shall both be rooted in social, political, and computer science theory as well as point in concrete directions, where solutions and opportunities in addressing these challenges may lie.
The workshop will, hence, produce:
- A much-needed vision on the role of AI and data analytics for the democratic role of the media.
- Best practices and insights into interdisciplinary engagement in value-sensitive algorithmic design of recommendation algorithms
- Guidelines as well as a manifesto for future research and long-term goals for the emerging topics of fairness, diversity, and personalization in recommender systems.
The Dagstuhl Perspectives Workshop 19482 on Diversity, Fairness, and Data-Driven Personalization in (News) Recommender Systems, took place from November 24 to November 29 at Schloss Dagstuhl in Germany. The goal of the workshop was to bring together researchers from the various disciplines relevant to news recommender systems (computer, communications, legal, and political science) to (1) develop a joint understanding of the issues arising for society with regards to the diversity and fairness of recommender systems, (2) identify the gaps in science, practice and regulation with regards to these topics, and (3) to compile a set of recommendations--in the form of a manifesto--that outlines needed steps from all actors involved to address the societal issues at hand.
The workshop was organized in the following phases:
Welcome and introductions This first phase introduced the workshop goal to the participants and then offered each of them five minutes to introduce their research activities, expertise, their interest in the topic, and research directions they see as relevant to the workshop's topic.
Impulse presentations Given the diversity of the backgrounds of the participants, eight brief stage setting presentations where given. The goal of these was to establish a common ground in terms of relevant questions and common vocabulary.
Topical breakout group discussions Based on the introducing presentations and impulse presentations, the next phase of the workshop was organized around topical breakout groups. Topics discussed included relating fairness to diversity, user desiderata and characteristics, wider societal implications, governance, data requirements, and clustering of research gaps.
Writing sessions The next phase was focused on jointly drafting the manifesto that incorporated recommendations developed from discussions so far and compiling them into a coherent document.
The remainder of this text provides the abstracts of the impulse presentations. The insights resulting from our discussions can be found in the manifesto document, which will be published in due course.
- Christian Baden (The Hebrew University of Jerusalem, IL) [dblp]
- Michael Beam (Kent State University, US) [dblp]
- Abraham Bernstein (Universität Zürich, CH) [dblp]
- Claes De Vreese (University of Amsterdam, NL) [dblp]
- Marc Hauer (TU Kaiserslautern, DE)
- Lucien Heitz (Universität Zürich, CH) [dblp]
- Natali Helberger (University of Amsterdam, NL) [dblp]
- Pascal Jürgens (Johannes Gutenberg-Universität Mainz, DE) [dblp]
- Christian Katzenbach (Institute for Internet & Society - Berlin, DE) [dblp]
- Benjamin Kille (TU Berlin, DE) [dblp]
- Beate Klimkiewicz (University Jagiellonski - Krakow, PL) [dblp]
- Wiebke Loosen (Universität Hamburg, DE) [dblp]
- Judith Möller (University of Amsterdam, NL)
- Goran Radanovic (MPI-SWS - Saarbrücken, DE) [dblp]
- Wolfgang Schulz (Universität Hamburg, DE) [dblp]
- Guy Shani (Ben Gurion University - Beer Sheva, IL) [dblp]
- Nava Tintarev (TU Delft, NL) [dblp]
- Suzanne Tolmeijer (Universität Zürich, CH) [dblp]
- Wouter van Atteveldt (VU University Amsterdam, NL)
- Sanne Vrijenhoek (VU University Amsterdam, NL) [dblp]
- Theresa Züger (Institute for Internet & Society - Berlin, DE)
- Katharina A. Zweig (TU Kaiserslautern, DE) [dblp]
- artificial intelligence / robotics
- data bases / information retrieval
- society / human-computer interaction
- Information filtering and recommender systems
- Algorithmic bias and data quality
- Balancing diversity and personalization
- Fair and transparent machine learning systems
- News personalization
- Political content and polarization
- Democratic theory