Personalized recommendation with multimodal explanations

The aim of this postdoctoral project are to design, develop and evaluate novel deep learning-based methods for multimodal product recommendation combining visual and textual product descriptions and metadata of a user. The main innovation regards the disentangled representation space in which preferences of users are matched with product attributes and that allow personalized, natural and multimodal explanations of why a product appeals to a particular user.

Keywords:Recommender system, Explainability, Multimedia mining, Personalization

Disciplines:Computer vision, Natural language processing, Interactive and intelligent systems, Pattern recognition and neural networks, Machine learning and decision making