Technologies for next-generation multi-media libraries: the contentus project

An ever-growing amount of digitized content as well as "born-digital" content published online forces libraries and archives to integrate new data sources and align them with their existing collections. This is a challenging task, since multimedia content is inherently diverse and integrating metadata is often complex due to the different structures, qualities and reliabilities of available sources. The CONTENTUS project investigates new solutions for libraries and multimedia archives.

In this talk, I will present our approach towards an integrated solution for libraries, archives and other content holders that facilitate a seamless transition from raw digital data to a semantic multimedia search environment. We aim to provide a complete set of tools ranging from quality control for digitization, content and metadata integration, to access through a semantic multimedia search and browsing interface.

The CONTENTUS multimedia search interface will offer integrated searches for texts, images, audio and audiovisual content in a unified semantic user interface. Search queries can be narrowed and expanded in an exploratory fashion, search results can be refined by disambiguating entities and topics, and semantic relationships become not only apparent, but can be navigated as well.

When the metadata available is not sufficient to describe the content for the purpose of the semantic search, CONTENTUS can generate the relevant metadata through a variety of content analysis techniques. These largely automated processing steps identify, among other things, named entities such as persons, places and organizations in texts, audio transcripts and audiovisual media. This makes it possible to associate content with matching authority file entries and ultimately to insert it into a growing semantic knowledge network that can then be searched or explored as described above.