Modeling Corporate Knowledge within the Agent Oriented Abstraction Pierre Maret and Jacques Calmet INSA de Lyon, France and University of Karlsruhe (TH), Germany E-mail: pierre.maret@insa-lyon.fr, calmet@ira.uka.de Abstract An enterprise can be considered in a distributed compu- tational paradigm. Multiagent systems have been invented to tackle such problems within artificial intelligence but are today seen also as a management methodology. The Agent Oriented Abstraction has been proposed to describe in a fullygeneric way a society of agents. In this paper, we are investigating the application of this model to the abstract modeling of corporate knowledge. We show that such an abstraction mechanism leads to very practical applications whether on the web or on any other support and that it cov- ers the broad scope of corporate knowledge. This approach can be further used to better simulate and support knowl- edge management processes. Key Words: Corporate knowledge, abstact modelling, agent soci- ety, agent oriented abstraction, knowledge management 1. Introduction The concept of corporate knowledge has at least two facets. A first one lies fully in business and refers to the amount and quality of know-how and information that is available throughout a company. The second one is set in the framework of information technology and is tightly linked to the various and diverse paradigms that are used to represent knowledge. Nowadays, with the common use of e-business these two concepts are becoming somewhat interrelated. Adopting the IT point of view leads to think of an enterprise as a distributed computational paradigm. Multiagent systems have been invented to tackle such prob- lems within artificial intelligence but are today seen also as a management methodology. It is thus not a new idea to model a company or enterprise as a multi-agent system. It is however a new idea to investigate what abstraction mech- anisms for agent methodologies can add to such a modeling. Work partially supported by the Calculemus Research Training Net- work HPRN CT-2000-00102; The work of Pierre Maret is partially spon- sored by the R´ egion Rh ˆ one-Alpes, France. In this paper we intend to investigate the application of the Agent Oriented Abstraction model [2] to the abstract modeling of corporate knowledge. Agent Oriented Abstrac- tion has been proposed to describe in a fully generic way the concept of agents. We show that such an abstraction mechanism leads to very practical applications for corpo- rate knowledge, whether on the web or on any other sup- port. The paper is organized as follows: section 2 gives an overview on existing approaches to corporate knowledge. Section 3 is a short presentation of the Agent Oriented Ab- straction. Section 4 consists of the description of the ab- stract modeling of corporate knowledge within the agent oriented abstraction model. Section 5 is a case study de- scription that serves to illustrate our approach. Our conclu- sions are presented in section 6. 2. Approaches to corporate knowledge Corporate knowledge generally covers several domains such as document/knowledge management, discussion fo- rums, skill management and knowledge engineering. The main paradigms to represent knowledge are logic, frames or semantic nets. Inference capabilities are closely associated to the paradigms of knowledge representation. The doc- ument/knowledge management domain is linked to docu- ment repositories (including Internet), indexing, search en- gines, information filtering, knowledge extraction through document mining just to name a few related fields. The- saurus and ontology are also widely used. Documents can be distributed, web-based, structured, semi-structured or unstructured. Discussion forums take advantage of commu- nication facilities within a community of individuals. The main concepts in this domain are topics, participants and messages. Recommender systems also known as collabora- tive filters [11] do belong to this domain. Skill management consists of identifying, recording and updating individuals competences in order to use them appropriately and to iden- tify lacks of competences [7]. Knowledge engineering con- centrates on capturing, structuring, inferring and distribut- ing knowledge. It is linked to specific knowledge engineer- ing techniques and to document repositories. Knowledge 1