Knowledge Embeddings for Explainable Recommendation

The explainable recommendation aims to create models that generate high-quality recommendations and intuitive explanations. The explanations may be ad hoc or derived directly from an explainable model (also called interpretable or transparent in some contexts). Explainable recommendation attempts to address the problem of why: by providing explanations to users or system designers, it assists humans in understanding why certain items are recommended by the algorithm, where humans can be either users or system designers. Explainable recommendations help recommendation systems improve transparency, persuasiveness, effectiveness, trustworthiness, and satisfaction. This work aims to use a knowledge embeddings extraction method to improve the explainability of recommender systems.