Conceptual Modeling in Multimedia Information Seeking [chapter]

Carlo Meghini, Fabrizio Sebastiani
1999 Lecture Notes in Computer Science  
Motivation Information retrieval, once considered "the Cinderella of computer science" (in the words of one of its main advocates), is now witnessing a booming interest in the light of the ever-growing amount of information repositories available on both local and, especially, distributed platforms, including the Web and the Internet at large. Other relatively new paradigms of information seeking, such as browsing, information gathering and information filtering, are just other tiles in this
more » ... eralised search for tools and techniques capable of reducing information overload and of selecting the right information at the right time in very large, dynamically evolving sets of documents. The availability of non-textual ("multimedia") documents has given a new twist to information retrieval research, unfortunately setting even farther in the future the time in which generalized, completely automatic indexing methods will be available that allow the answering of content-based queries. If a system really capable of fully automatic recognition of "Authoritative sources on the social impact of nuclear waste disposal" seem currently out of reach, even more so for needs of "piano sonatas in styles influenced by art nouveau", "cubist paintings representing violins" or "scenes on sieges from war movies". Complex queries such as the above mentioned only reflect the complexity of the information needs of many sophisticated information seekers of today; and it is only apparently that the queries of less sophisticated users have a smaller intrinsic complexity. While in a few decades these information needs might be satisfied by fully automatic systems, for today and tomorrow we will have to make do with systems relying on a mixture of automatic, semi-automatic and manual indexing methods, as it is only through manual intervention that it is possible to inform a retrieval system of the ins and outs of a (non-textual) document. In this respect, an important contribution to multimedia information seeking may come from the more content-minded conceptual modelling and knowledge representation communities, as these have been developing formal and semi-formal languages and methodologies for the representation of information and for the conceptualization of an application domain. It is expected that, in the mid to long term, successful information seeking systems will have to be based on the close interplay of manually (or semi-automatically) created representations of document content based on conceptual modelling technology, and the automatically created representations of document form of the information retrieval tradition. Topics addressed at the workshop The papers presented at the Workshop address several related aspects of the information seeking problem. In the following, we have categorized them in three broad classes: modeling Web documents, modeling multimedia documents, and browsing. While the last class is clearly disjoint 1
doi:10.1007/3-540-48854-5_22 fatcat:qnlf4dojdbeyzj7mfbsxi2o6uy