Tweet location inference based on contents and temporal association

Abstract

How can we infer a tweet location? Are timestamps of tweets effective for the location inference? In this study, we propose a novel method for tweet location inference based on contents and timestamps of tweets. It is important to infer the locations of tweets for the services related to locations such as recommending restaurants, sending disaster-related information to users, and providing commercial messages to users. This study has two contributions: (1) we propose a novel method to infer tweet locations based on the contents and timestamps of tweets, andbreak (2) we experimentally demonstrate the effectiveness of the proposed method using Twitter data. The experimental results suggest that the proposed method can infer tweet locations more precisely than a baseline that does not take the temporal association into account.