Towards a Character-based Meta Recommender for Movies

Abstract

Improving user experience through personalization is one of the current research trends in HCI. This includes recommendations that suit the preferences of all users (not only the majority) as dictated by their character, i.e. all aspects that influence human behaviors; including personality traits, affective states, socio-cultural embeddings and individual beliefs to name but a few. The aim of this paper is developing a recommender system for movies, that is adaptive in the way it recommends selections on the basis of the user’s character. We present an architecture for a generic module-based recommendation platform that uses the user’s character to choose a recommendation algorithm for each user. We deployed a movie recommendation application to determine the relation between recommender algorithm preference and (1) the user’s personality, (2) background and (3) gender. Based on the data collected from 84 participants, high correlation between the user’s personality (Openness, Extraversion and Conscientiousness) and gender and the recommender algorithm that they prefer.