A critical approach on reusing ontologies

abstract: The purpose of this paper is to summarize what types of problems occur when building ontologies with a maximum reuse of existing pieces. That is, not the kind of problems occuring for ex-nihilo knowledge modelling, but rather for expanding a core ontology about a spe-ciic domain with parts of semantically overlapping ones. The types of problems we raise cover three areas: Importing ontologies, meta-operators for ontology generation, and domain representation complexity. We rst deene ontologies relatively to other knowledge representation terms, and describe operations we need in order to reuse ontologies. Instead of setting an expression formalism, we suggest the use second order operators as type constructors to generate a formal language appropriate for a given organisational level. On a \toy example", we notice that obtained domain knowledge is rather large. We use layering ontologies as a mean to contain complexity on each organisationnal level. acts, robotics This paper has not already been accepted by and is not currently under review for a journal or another conference. Nor will it be submitted for such during IJCAI's review period. 1 1 About ontologies 1.1 Deenitions In this part we will summarize deenitions of ontology and related terms. According to these deenitions we may submit a classiication for ontologies. Ontology : The notion of ontology is deened by Tom Gruber Gru93] as a speciication of a conceptualization. That is, an ontology can be seen as a set of concepts and relations between these concepts, that describes a speciic part of knowledge. Since an ontology is designed by a given agent, or community of agents, in order to describe some piece of knowledge, it cannot be considered as an extensive description, but rather as a point of view on that piece of knowledge. Domain knowledge : Domain knowledge, also called domain theory, speciies the structure of a domain speciic knowledge that can be useful for an application. The domain is rather speciic (for instance, well-known examples of domain knowledge are elevator design or ight assignment) and therefore, domain theory will be highly application de-pendant. Meta knowledge : Meta knowledge express the structure underlying the domain knowledge (i.e sets, numbers...) and is not supposed to depend from any application. Therefore, it is highly generic and reusable. Meta knowledge, as presented in CommonKADS BV94] includes ontolo-gies, case model (task conceptualization, recursive division in sub-tasks) and problem solving methods (paradigms employed for solving an elementary …