ERCIM News No.41 - April 2000 [contents]

Corporate Knowledge Management through Intranet and Internet

by Rose Dieng

The ACACIA research team at INRIA-Sophia-Antipolis, aims at offering methodological and software support in form of models, methods and tools for knowledge management ie, for building, managing and distributing a corporate memory.

The ACACIA team studies the problems raised by the dissemination of knowledge through a knowledge server via the company’s Intranet or via the Web: we consider the Web as a privileged way for supporting management of knowledge distributed either inside a company or between several companies (for example in the framework of a project memory or of technological monitoring). We aim at building knowledge servers enabling search for information in a heterogeneous corporate memory, this search being ‘intelligently’ guided by ontologies and ontological annotations. We focus on the case of a corporate memory materialised in the form of XML documents. The ontologies can then be expressed either in RDF Schema, or in Sowa’s Conceptual Graphs or in CML (CommonKADS Modelling Language).

Research on Knowledge Servers

We developed WebCOKACE, a knowledge server for distributing CommonKADS expertise models on the Web. It enables to explore expertise models and to perform hypertextual navigation in expertise models and towards associated electronic documents thanks to a Web browser .

Research on XML Galaxy

Considering that information search in the memory can be guided by knowledge models and semantic metadata, Acacia studies information search in XML documents semantically annotated by RDF metadata or by conceptual graphs. We built a translator transforming an ontology expressed in CommonKADS, into an RDF schema that may serve as an annotation model of the corporate XML documents. We built an ontological filtering engine to answer simple requests of the user. We proposed a model of interpretation of RDF and RDF Schema in terms of conceptual graphs. We developed a translator of RDF Schema/RDF towards conceptual graphs. It enables to exploit projection on conceptual graphs in order to retrieve the adequate XML documents semantically annotated by RDF metadata thus translated into conceptual graphs. This work is implemented in JAVA in the OSIRIX tool. We are studying exploitation of complex inference mechanisms on conceptual graphs in order to enable different kinds of approximate search of information.

We also developed an inference engine able to perform elementary, event-guided, processings, on XML documents: consistency management, elementary deductions. This engine was applied to make deductions using a XML-expressed knowledge base in road accident analysis.

Moreover we are studying an exploitation of RDF and RDF Schema for taking into account the notion of viewpoint. The purpose is to enable to represent an ontology with multiple viewpoints, corresponding for example to multiple professions in a company.


We developed RESEDA, a system aimed at supporting road accident analysis. Installed on the intranet of INRETS, it enables the investigators of this institute to fill the accident dossiers in a computational form, with an intelligent support. The check-lists constituting an accident dossier are reconstituted from a generic XML document and from data stored in a database. According to the data entered about the current accident, RESEDA exploits a knowledge base (written in XML format) in order to provide the user with suggestions useful for his task of accident analysis. These suggestions stem from a knowledge base that was built thanks to acquisition and modelling of knowledge of experts of INRETS. RESEDA also exploits a base of generic scenarios (written in XML format) in order to suggest the most plausible generic scenarios that may be applicable for the reconstitution of the current accident analysed by the user. The scenarios are visualized thanks to XSLT stylesheets. RESEDA illustrates an application combining knowledge engineering and XML-based technologies.

Cooperative projects

The ACACIA project takes part in two IST projects that started officially in February 2000:

The ACACIA team also takes part in the cooperative project ESCRIRE (Embedded Structured Content Representation In REpositories) with EXMO and ORPAILLEUR, two other teams of INRIA (see articles on pages 9 and 10). This project aims at comparing three types of knowledge representation formalisms (conceptual graphs, object-based knowledge representation and description logics) from the standpoint of the representation and the handling of document content.

Last, the Acacia project takes part in the Working Group on Agent-Based Mediation Among Information Systems, Data, and Knowledge Management, in the framework of the AgentLink SIG on Intelligent Information Agents.

ACACIA website: http://www.inria.fr/Equipes/ACACIA-eng.html
WebCokace: http://inria.fr/acacia/Cokace

Please contact:
Olivier Corby and Rose Dieng - INRIA
Tel: +33 4 9238 7871/7810
E-mail: Olivier.Corby@sophia.inria.fr, Rose.Dieng@sophia.inria.fr