A Japanese-Chinese Cross-Language Entity Linking Method Based on Appropriateness of Term Description Pages

In this paper, we propose a method to automatically discover valuable key phrases in a Japanese document and to link these key phrases to corresponding Chinese Wikipedia pages. The method that we propose has three stages. First, we translate Japanese key phrases into Chinese using combination of three translation methods. Second, we extract all the Chinese Wikipedia articles of the translated key phrases. Third, we translate the original Japanese document into Chinese and make a vector of noun frequencies. We calculate the cosine similarity of the original and all of the candidate Chinese Wikipedia articles. To find the appropriateness of term description pages for disambiguation, we make a ranking using cosine similarity comparison of the Japanese document and Chinese Wikipedia articles. Finally, we make a link from the Japanese key phrase to the top ranked Chinese Wikipedia article.