Exploiting distributional semantics models for natural language context-aware justifications for recommender systems

In this paper we present a methodology to generate context-aware natural language justifications supporting the suggestions produced by a recommendation algorithm. Our approach relies on a natural language processing pipeline that exploits distributional semantics models to identify the most relevant aspects for each different context of consumption of the item. Next, these aspects are used to identify the most suitable pieces of information to be combined in a natural language justification. As information source, we used a corpus of reviews. Accordingly, our justifications are based on a combination of reviews' excerpts that discuss the aspects that are particularly relevant for a certain context. In the experimental evaluation, we carried out a user study in the movies domain in order to investigate the validity of the idea of adapting the justifications to the different contexts of usage. As shown by the results, all these claims were supported by the data we collected.