Recommendations for the Recommenders: Reflections on Prioritizing Diversity in the RecSys Challenge
Open Access
- 14 October 2024
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 22-26
Abstract
The RecSys Challenge 2024, co-organized by the Danish news outlet Ekstra Bladet, called for participants to develop a news recommender system that accurately predicts which news article a reader is most likely to select from a given list. In the context of this challenge, the organizers explicitly stated their interest in beyond-accuracy objectives. However, the setup of the challenge did not facilitate pursuing these more normative goals: a missed opportunity, given the quality of the dataset. In this paper, we highlight the issues encountered in a submission that prioritized normative diversity. We reflect on the responsibility of conferences, RecSys in particular, when it comes to promoting beyond-accuracy objectives and provide recommendations for future challenge iterations.Keywords
This publication has 25 references indexed in Scilit:
- Intra-list similarity and human diversity perceptions of recommendations: the details matterUser Modeling and User-Adapted Interaction, 2022
- Codabench: Flexible, easy-to-use, and reproducible meta-benchmark platformPatterns, 2022
- Recommended for You: How Newspapers Normalise Algorithmic News Recommendation to Fit Their Gatekeeping RoleJournalism Studies, 2022
- Benefits of Diverse News Recommendations for Democracy: A User StudyDigital Journalism, 2022
- Recommenders with a Mission2021
- Beyond Optimizing for Clicks: Incorporating Editorial Values in News Recommendation2020
- Are we really making much progress? A worrying analysis of recent neural recommendation approaches2019
- RecSys Challenge 20172017
- Diversity, Serendipity, Novelty, and CoverageACM Transactions on Interactive Intelligent Systems, 2016
- Improving Aggregate Recommendation Diversity Using Ranking-Based TechniquesIEEE Transactions on Knowledge and Data Engineering, 2011