Ranking Interaction-Based Collaborative Filtering Recommendations Using Temporal Features In Online Dating

Users of social networks such as dating websites need some help to find their successful matches. Most of the existing recommenders use either profile similarity or interaction similarity to recommend new matches. However, temporal features are not being used in these recommenders. This paper discusses the results of a temporal data analysis experiment using a dating website's data. Then the results of a recommender system that uses temporal features with a collaborative filtering recommender will be shown. Although the improvement of the recommendations after using the temporal features is very small, the post-experiment and pre-experiment data analysis suggests that temporal features can improve the recommendations further.