![]() ![]() ![]() |
![]() |
|
|
![]() ![]() ![]() ![]() ![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Return to Papers In this paper, we propose an integrated retrieval system that retrieves distributed multimedia information from existing heterogeneous information sources by entering any media key independent of the sources. The main feature of the proposed system is a media translation dictionary that can translate the medium of the retrieval key entered into media appropriate for the available information sources by connecting different media data. Furthermore, it issues query statements compatible with each heterogeneous information source to retrieve distributed multimedia information. In addition, the dictionary can be automatically made from learning data. We implement a prototype system and perform image and text integration retrieval experiments. The effectiveness of making the dictionary automatically is confirmed. ![]() DiSC'02 © 2003 Association for Computing Machinery |