"Where can I Buy iPhone4S Now?": Spatio-Temporal Entity Retrieval on Twitter
Yan Liang, Qiang Ma, Masatoshi Yoshikawa · 2012
Entity retrieval is a task to find information related to real-world-entities (person, organization. location. product, etc.). In this paper, we focus on the issue of spatio-temporal entity retrieval on Twitter, which searches for entities related to a given entity at a certain time in a certain region. For a given query, like where can I buy iPhone4S now, we capture the ERE (Entity-Relation-Entity) tuples from tweets and compute their scores to find the related entities (shops in our example) by considering: 1) terms which describes relations between entities or status and behaviors of entities, 2) properties of twitter users, 3) sentence structure, 4) descriptive polarities on entities and 5) time and location of the user who issued that query. We have designed the experiments for verifying the proposed method. Keyword Entity Retrieval,Relation Mining,Spatio-Temporal Retrieval, Twitter