Improving Case-Based Recommendations using Implicit Feedback

摘要

A recommender system suggests items to a user for a given query by personalizing the recommendations based on the user interests. User personalization is usually done by asking users either to rate items or specify their interests. Generally users do not like to rate items; an alternative approach would be to implicitly track user's behaviour by observing their actions. In this paper, we build a recommender system by using case-based reasoning to remember past interactions with the user. We incrementally improve the system recommendations by tracking user's behaviour. User preferences captured during each interaction with the system are used to recommend items even in case of a partial query. We demonstrate the proposed recommender system in a travel domain that adapts to different kinds of users.