Self-explaining recommendation

Abstract

[en] Recommender Systems are increasingly becoming more and more part of our everyday life, from helping us choose what videos to watch to recommending news articles and friends. However, their architecture is really complex which makes it difficult to explain their suggestions. The goal of this project is to study in depth how does a Recommender System work, which generates automatically, in natural language, explanations on their recommendations. Furthermore, its details will be studied with different settings. It will be evaluated, analysed and discussed with different datasets.