インターネット動画の単語検索の向上

摘要

A large number of vides files are shared in video sharing services in the Internet. In this paper, we propose a video community extraction method based on web community extraction methods, which consist of fan pages and center pages, and TF-IDF. With our proposed method, a video and a video list are considered as a center page and a fan page in web community extraction, respectively. For evaluation, we executed video search based on the proposed video community extraction and compared its search results with those using other methods. Our experimental results have shown that the proposed method is able to provide search results better than other methods.%インターネット上の動画共有サービスが普及し，多くの動画が動画共有サイトで共有されている.しかし，動画共有サイトやWeb検索エンジンで提供されている動画の単語検索機能では検索語と関連した単語を得ることは容易ではなく，その精度は必ずしも十分とは言えない.よって，動画共有サイトにおける単語による動画検索精度の向上は重要な課題のーつと考えることができる.