Visual analysis for semantic search in digital libraries

Semantic search of cultural content is of major importance in current cultural digital libraries, such as Europeana, or the evolving Digital Public Library of America. Content metadata accompanying the digitised items are analysed, mapped and used to interpret users' queries, so that the most appropriate content is selected and presented to them. Multimedia, especially automatic visual analysis, has not been a main component yet. This paper presents a semantic search methodology, including a query answering mechanism which meets the semantics of users' queries and enriches the answers by exploiting appropriate local (SURF) and global (MPEG-7) visual features and descriptors. An experimental study is presented, using content from the Europeana digital library, involving both thematic knowledge and visual analysis of cultural images, illustrating the improved content search performance. (4 pages)