Extraction of Location related Terms Using Query Logs

Abstract

In this paper, we propose a method to extract location-related terms (categories) with query logs. There are some terms co-occurring with location names in query logs, and we think the terms represent interests of the location. However, it is difficult to grasp trend of interests of the location, because the kind of terms appear in large amounts. Even if they ware restricted a few frequent terms, you can only grasp one part of interests of the location. So we solve this problem using location-related terms (categories). Our method has four processes, (1) extracting terms co-occurring with location name in query logs, (2) clustering the terms, (3) naming the each cluster set location-related term (category), (4) in each locations, calculating percentage of location-related terms using frequency of co-occurrence terms. In the experiment, we show that our method is useful to grasp the interests of the location and comparison with interests of others.