Semantic information brokering: how can a multi-agent approach help?
CIA'99 Proceedings of the 3rd international conference on Cooperative information agents III
Hi-index | 0.00 |
Visual information, especially videos, plays an increasing role in our society for both work and entertainment as more sources become available to the user. Set-top boxes are poised to give home users access to videos that come not only from TV channels and personal recordings, but also from the Internet in the form of downloaded and streaming videos of various types. Current approaches such as Electronic Program Guides and video search engines search for video assets of one type or from one source. The capability to conveniently search through many types of video assets from a large number of video sources with easy-to-use user profiles cannot be found anywhere yet. VideoAnywhere has developed such a capability in the form of an extensible architecture as well as a specific implementation using the latest in Internet programming (Java, agents, XML, etc.) and applicable standards. It automatically extracts and manages an extensible set of metadata of major types of videos that can be queried using either attribute-based or keyword-based search. It also provides user profiling that can be combined with the query processing for filtering. A user-friendly interface provides management of all system functions and capabilities. VideoAnywhere can also be used as a video search engine for the Web, and a servlet-based version has also been implemented.