Relevance Analysis among Domestic Enjoyments based on News Site App Users' Interest

Abstract

We proposed a method to discover domestic enjoyments information that are specific and original to individual domestic area. The proposed method is based on article follow information of news site users and themes given to articles. We extracted the relationship between regional theme and sightseeing theme using co-occurrence information of themes in each article. In addition, We pulled out the relationship between regional theme and sightseeing theme using themes actively selected by the user for personalization. By filtering with similarities between those two types of relevance, accuracy improvement was achieved. Moreover, by focusing on sightseeing theme with low similarity, error candidates for theme linking can be extracted.