News Curation Service using Semantic Graph Matching
Ryohei Yokoo, Takahiro Kawamura, Yuichi Sei, Yasuyuki Tahara, Akihiko Ohsuga · 2015
In recent years, Curation Services that rec- ommend on the Internet to users are getting attention. In this paper, we propose a curation service that collects and recommends news articles that users feel by using semantic relationships between terms in the articles. We define interested as that users have curiosity and serendipity. The semantic relations between events terms are represented by Linked Data. We create News Articles Linked Data (candidates for recommendation to users) and User's preferences Linked Data (users' preferences). In order to recommend to users, we first search common subgraphs between two kinds of Linked Data. The experiment showed that the curiosity score is 3.30 (min:0, max:4), and the serendipity score is 2.93 in our approach, although a baseline method showed the curiosity score is 3.03, and the serendipity score is 2.79. Thus, we confirmed that our approach is more effective than the baseline method. Keywords-Semantic Relation; Linked Data; News Recommen- dation.