Loosely coupled interest groups, so-called communities of interest, are valuable sources of knowledge, which have become accessible to large audiences via the Internet. The exchange of documents via document catalogs is an important means of knowledge exchange among community members. The individual community members often organize personal document collections in personal catalogs, such as for example bookmarks. So far, the additional effort that community members have to invest for knowledge exchange, for example via a community catalog, strongly impedes knowledge exchange in communities. Our first goal in this work is to design an application framework for supporting knowledge exchange via document catalogs in communities. We analyze requirements for the framework, based on psychological and sociological characteristics of the knowledge exchange process in communities. We introduce the CAIMAN framework, which consists of knowledge exchange support services that fulfill the above requirements and a mediation infrastructure with which the services can be realized. CAIMAN allows a community member to use one personal document catalog for all knowledge exchanges, independent the catalog choice of the exchange partner. The exchange among the different catalogs is delegated to the CAIMAN services and mediation infrastructure. Our second goal is to provide and evaluate a concept for a document catalog mediation infrastructure. Our mediation approach provides for semi-automatic virtual integration of document catalogs. Conceptual catalog differences are resolved by a catalog matching component and issues of querying heterogeneous document catalogs are taken care of by a catalog query infrastructure. We introduce a novel catalog matching approach, which automatically matches the categories of two catalogs. Our matching approach uses the documents and the structure of a catalog for the matching calculations, whereas the highly subjective category names are not used. The CAIMAN matching approach is based on text classification and graph matching techniques. Our querying approach allows to jointly query heterogeneous document catalogs that are represented with different data models, using the same query infrastructure. The approach requires the conversion of all catalog data to RDF (Resource Description Framework) as lingua franca for catalog querying. As a proof of concept for our querying approach, we show how Topic Maps based document catalogs can be jointly queried with RDF based document catalogs. RDF and Topic Maps are the most popular catalog representation data models on the Web today. We have implemented a catalog mediation prototype. Experimental results show that our approach provides for high-quality mediation, which is the basis for successful knowledge exchange.
«
Loosely coupled interest groups, so-called communities of interest, are valuable sources of knowledge, which have become accessible to large audiences via the Internet. The exchange of documents via document catalogs is an important means of knowledge exchange among community members. The individual community members often organize personal document collections in personal catalogs, such as for example bookmarks. So far, the additional effort that community members have to invest for knowledge e...
»