タイトル:		Semantic analysis of a large-scale news video archive
著者：			Ichiro Ide, Akira Ogawa, Kazuhiro Noda, Shin'ichi Satoh, Hiroshi Murase
所属：			Nagoya University / National Institute of Informatics
アブストラクト：
We introduce our recent works on semantic analysis on a large volume of  
news video data archived for more than five years, equivalent to  
approximately 900 hours of MPEG-1 video data.
After briefly introducing the archive, two works that analyzes the news  
contents based on text and image information is introduced; topic  
threading and cross-lingual news story retrieval.
At the end is introduced an important image processing method for the   
semantic analysis; fast near-duplicate video segment detection.
Some of the works introduced in the presentation are still in a  
preliminary stage, but we believe that they should play an important role  
in handling the growing amount of video contents in the future.