XRecSys: A framework for path reasoning quality in explainable recommendation

There is increasing evidence that recommendations accompanied by explanations positively impact on businesses in terms of trust, guidance, and persuasion. This advance has been made possible by traditional models representing user–product interactions augmented with external knowledge modeled as knowledge graphs. However, these models produce textual explanations on top of reasoning paths extracted from the knowledge graph without considering relevant properties of the path entities. In this paper, we present XRecSys, a Python framework for the optimization of the reasoning path selection process according to properties deemed relevant by users (e.g., time relevance of the linking interaction or popularity of the entity linked to the explanation). Our framework leads to a higher reasoning path quality in terms of the considered properties and, consequently, textual explanations more relevant for the users.