acm-header
Sign In

Communications of the ACM

ACM TechNews

Video Search Solves the News Segmentation Problem


View as: Print Mobile App Share:

Computer scientists at Paris-based search engine company Exalead have developed Voxalead, a video search service that automatically divides news broadcasts into standalone passages on specific topics.

Voxalead can produce an automated transcript of what is being said in news videos and then allow for a search with keywords in a straightforward way. Voxalead makes use of software that analyzes transcripts for obvious transitions from one segment to another by looking for repetitions of nouns, adjectives, and certain verbs, and then testing the various patterns of segmentation.

"The general idea of this method is to search for the best possible segmentation among all the possible ones," say the computer scientists.

They say the approach provides accurate results, allows users to search for events, and can show timelines of the number of topics mentioned in the past. It also offers a geographical mashup that indicates where events took place and a tag cloud showing trending topics.

"The system processes more than 150 new video and audio items each day, amounting to roughly 3.5 GB or 15 hours of new daily content, on a single six-core server," notes the team, which means it could process about 100 hours of video a day.

From Technology Review
View Full Article

Abstracts Copyright © 2011 Information Inc. External Link, Bethesda, Maryland, USA 


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account