Technical Workshop on Information-Management in Sciences
Department of Physics,
Carl von Ossietzky Universität Oldenburg, Germany
11 - 13 October 1999
 
CRISP II
Home
Programme
 
Proceedings
 
Participants
 
Registration
 
Travel &
Housing
Links
 
Contact
Last conference CRISP'97
 
     

    Thomas Severiens

    CARMEN

    Abstract: The main focus of the work in the CARMEN project lies in the further development of content analysis by means of new procedures with strong links to retrieval.

    Efforts to produce homogeneity and consistency in today's decentralised world of information rely on the creation of suitable information systems for distributed collections of data. Often technologically oriented solutions are sought for the problems so that it is possible to access simultaneously different document pools. This is however insufficient in itself. The main problem of both the differences in content and the conceptual differences between the various individual collections of data has not yet been resolved. For this reason, in this proposal new solutions to the problems as well as further developments are scheduled in the following three areas:

    • Metadata.
    • Methods of dealing with the remaining heterogeneity.
    • Retrieval for structured documents with metadata and heterogeneous datatypes.
    The first two of these areas are closely connected. Through the further developments in the field of metadata the lost consistency should be partially re-established. With the methods of dealing with the remaining heterogeneity documents with varying levels of data relevance and content analysis should be treated in the sense of a layer model. On the other hand, these two areas should be supplemented from the retrieval side, similar to the work of a pair of tongs, by means of a search strategy which takes into account the different types of data with varying formats of metadata and the strong textual structuring (XML formats). At the same time, various aspects of search have to be considered, such as hypertext browsing, full-text search or related document search. Neither current hypertext nor commercially available text-retrieval systems provide this, and that is why further developments are necessary. These will include existing systems as far as possible (integration of Harvest).

    Prototypical installations shall make the progress visible and evaluable which has been achieved in various subprojects of CARMEN.

    Slides: CARMEN
    Links:
  • Global-Info
  • last update: 15. Feb. 2008   e-mail CRISP II