MindReader

MindReader is a novel dataset providing explicit user ratings over a knowledge graph within the movie domain. The latest stable version of the dataset contains 218,794 ratings from 2,316 users over 12,206 entities entities, and an associated knowledge graph consisting of 18,133 movie-related entities. The dataset is collected from an online movie recommendation game, MindReader, where users are pseudo-randomly asked to provide preferences for both movie- and non-movie entities (e.g., genres, actors, and directors). For each entity, users can either like it, dislike it, or state that they do not know it.