Daisuke Ikeda, Toshiaki Fujiki, Manabu Okumura
People often write in their blogs about news articles or events in news articles. In this case, however, the details of the news articles or events are often poorly described in such blog entries. Therefore, the readers of blogs need to find the original articles, which contain more details of the news articles, when they want to know about them. In this paper, we propose a method for linking news articles to blog entries that refer to them. Since blog entries and news articles are considered to be rather different, the common model for linking is not applicable. Therefore, we similarly used a vector space model for finding similar documents, but we tried to devise a new weighting method and a distance metric to improve the performance, by taking into account the properties of blog entries and news articles document sets.