A Text-based Method for Detection and Filtering of Commercial Segments in Broadcast News
Ganesh Ramesh (Department of Computer Science, University at Albany, SUNY, 1400, Washington Avenue, Albany, NY 12222, USA.)
Amit Bagga (Avaya Labs Research,233 Mt. Airy Road, Basking Ridge, NJ 07920,USA.)
WO24: Applications Based On Written LRs
Story segmentation is an important problem in multimedia indexing and retrieval and includes detection of commercials as one of its component problems. Commercials appear regularly in television data and are usually treated as noise. Hence, filtering of commercials is an important task. This paper presents a system that detects and filters commercials from broadcast news data. While previous work in the area relies largely on features from audio, video and captions, the system described in this paper uses just closed caption text to perform this task. An evaluation of this system is also presented which shows comparable performance with other methods.
Closed caption text, Broadcast news, Keyphrase, Commercials, Story segmentation