The Exploration-Exploitation Trade-off in Interactive Recommender Systems
Author:
Open Access
- 27 August 2017
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 431-435
Abstract
Recommender Systems (RS) help users discover interesting products by means of relevant proactive suggestions. To accomplish this, RS must learn about user's unknown/unclear tastes, and constantly adapt to the dynamic nature of their environment. In this research, I focus on the notion that RS are faced with an optimization problem when generating recommendations: exploit the known user model, or explore other preferences the user might have. My PhD work aims to define the role of exploitation and exploration in RS, and proposes mechanisms that would allow to balance and control this trade-off. In this extended abstract I present the motivation, related work and research plan that guide the project.Keywords
Funding Information
- Science Foundation Ireland (SFI/12/RC/2289)
This publication has 13 references indexed in Scilit:
- Diversity, Serendipity, Novelty, and CoverageACM Transactions on Interactive Intelligent Systems, 2016
- Recommendations with a Purpose2016
- Adaptive, Personalized Diversity for Visual Discovery2016
- Recommender Systems Handbook2015
- Exploration in Interactive Personalized Music RecommendationACM Transactions on Multimedia Computing, Communications, and Applications, 2014
- A contextual-bandit approach to personalized news article recommendation2010
- Learning and adaptivity in interactive recommender systems2007
- Being accurate is not enough2006
- Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensionsIEEE Transactions on Knowledge and Data Engineering, 2005
- Reinforcement Learning: An IntroductionIEEE Transactions on Neural Networks, 1998