Organizational Semantic Webs

IntroductIon The main subject tackled in this article is the use of knowledge technologies to develop corporate memories or (stated more generally) " organizational memories " (OMs) (Dieng, Corby, Giboin, & Ribière, 1999). At the end of the 1990s, AI technologies, in general , and knowledge technologies, in particular, were recognized as pertinent and promising tools (in addition to information technologies) for the de-These very diverse technologies (concepts, methods, and tools) have been conceived to assist knowledge acquisition, modeling, and discovery, as well as the development of knowledge-based systems (Studer, Benjamins, & Fensel, 1998). In this article, we focus on knowledge modeling and formalization techniques, since our prime interest is the preservation of knowledge within OMs and its impact on the exploitation of this knowledge. In practice, the use of these technologies generates two complementary proposals: (1) the for-malization of a part of knowledge to be preserved, which means considering hybrid memories in terms of specification modes (formal, semi-formal, and informal); and (2) the introduction of a formal ontology of the domain in question, in order to facilitate the expression, comprehension, and access to capitalized knowledge. Formalization thus relates to both (1) knowledge (as proposi-tional knowledge) and (2) meaning (as conceptual knowledge). Regarding the balance between formal and informal specification, a broad spectrum of OM architectures have been proposed, ranging from informal annotation of formal knowledge bases (Euzénat, 1996) to the formal annotation of informal documents (Buckingham Shum, Motta, & Domingue, 2000). It should be noted that these extremes (i.e., the development of a text-documented knowledge base and the publication of scientific articles on the Web, respectively) correspond to atypical OM applications. The knowledge technologies used in 2004 to develop OMs are generally those of the Semantic Web, where languages like OWL (Antoniou & van Harmelen, 2004) allow us to exchange knowledge bases on the Web. One particular asset of OWL is its ability to offering several dialects with different expressive powersthe choice of the dialect depending on the specific application in question. A review of the state of the art (cf. section 2) shows that current OM architectures rely on " lightweight " knowledge models, corresponding to formal annotations of textual resources. These approaches focus on document " enrichment " (Motta, Buckingham Shum, & Domingue, 2000), since the knowledge models and ontologies are used to facilitate access to textual resources and the dissemination of the latter to interested users. …