Many online services provide users with recommendations, to help them find items of interest in the enormous space of available choices. Popular examples of such recommender services include videos (YouTube), music (Spotify), movies (Netflix), online shopping (Amazon), etc. These services typically base their recommendations on items liked/disliked by users.
In this project we address a narrative-driven recommendation scenario (detailed in this paper), where the user specifies her preferences in a short textual summary, complemented with a small set of positive and negative examples. The objective is to develop novel recommendation methods that can effectively utilize this information.