Relating Newcomer Personality to Survival and Activity in Recommender Systems

In this work, we explore the degree to which personality information can be used to model newcomer retention, investment, intensity of engagement, and distribution of activity in a recommender community. Prior work shows that Big-Five Personality traits can explain variation in user behavior in other contexts. Building on this, we carry out and report on an analysis of 1008 MovieLens users with identified personality profiles. We find that Introverts and low Agreeableness users are more likely to survive into the second and subsequent sessions compared to their respective counterparts; Introverts and low Conscientiousness users are a significantly more active population compared to their respective counterparts; High Openness and High Neuroticism users contribute (tag) significantly more compared to their counterparts, but their counterparts consume (browse and bookmark) more; and low Agreeableness users are more likely to rate whereas high Agreeableness users are more likely to tag. These results show how modeling newcomer behavior from user personality can be useful for recommender systems designers as they customize the system to guide people towards tasks that need to be done or tasks the users will find rewarding and also decide which users to invest retention efforts in.