Trip Tweets Search by Considering Spatio-temporal Continuity of User Behavior

A large amount of tweets about user experiences such as trips appear on Twitter. These tweets are fragmented information and not easy to share with other people as a whole experience. In this paper, we propose a novel method to find and organize such fragmented tweets at the level of user experiences. The notable feature of our method is that we find and organize tweets related to a certain trip experience by considering the spatio-temporal continuity of user-behavior of traveling. First, we construct a co-occurrence dictionary by considering the spatio-temporal continuity; i.e., the co-occurrence ratio of two terms is varying in time scopes and regions. Then, we use such dictionary to calculate the relatedness of a tweet to the trip experience from three aspects: content relatedness, temporal relatedness, and context relatedness. Tweets with high relatedness scores will be returned as search results. The experimental results showed our method performs better than conventional keyword-based methods.