Knowledge Analysis on Process Models

Abstract

Helping end users build and check process models is a challenge for many science and engineering fields. Many AI researchers have investigated useful ways of verifying and validating knowledge bases for ontologies and rules, but it is not easy to directly apply them to checking process models. Other techniques developed for checking and refining planning knowledge tend to focus on automated plan generation rather than helping users author process information. In this paper, we propose a complementary approach which helps users author and check process models. Our system, called KANAL, relates pieces of information in process models among themselves and to the existing KB, analyzing how different pieces of input are put together to achieve some effect. It builds interdependency models from this analysis and uses them to find errors and propose fixes. Our initial evaluation shows that KANAL was able to find most of the errors in the process models and suggest useful fixes including the fixes that directly point to the sources of the errors.