Discussion and Conclusion

SeSKA (Seamless Structured Knowledge Acquisition) is a methodology that is supported by a graphical tool SOOKAT (Structured Object-Oriented Knowledge Acquisition Tool) that is used to create iteratively object-oriented models of three types: the domain model (DM), the dependency graph (DG) and the inference model (IM) containing the inference structure (IS).

Inferences can be performed in instantiations of the models. The DM and DG are formed from small fractions of knowledge, the IM based on the DG. The DM can also be formed with the help of statistical analyses used semi-automatically for forming a partial formal grammar.

When viewing SOOKAT tool from the ontological point of view, it is discovered that its models are different kinds of ontologies. The domain model is a subset of a domain ontology, the DG is an application ontology, and the IS is a task ontology.

The graphical features of SOOKAT enhance terminology management during different phases of KA.

Similarity analysis of texts have bee applied also elsewhere, but using a different approach.

The main contributions of this paper are

• applying statistical text analysis techniques to domain conceptualization in knowledge acquisition, i.e. to forming a domain ontology,
• discovering that models of the tool SOOKAT are different kinds of ontologies, the formation of which can be described and
• suggesting options about terminology management and integration.

Future directions of research include developing the tool SOOKAT as well as further testing of statistical methods in constructing ontologies.

Sat Oct 14 22:52:14 EEST 2000