Extracting Constraint Knowledge from Code: A Case-Based Reasoning Approach

Abstract

The underlying motivation for the authors' research on constraint understanding is comparative constraint understanding. The same constraint can be expressed in different ways in different models, and a comparative constraint-understanding system would be able to assess the extent to which two models have equivalent constraint semantics. Understanding constraint semantics can be used to support tasks such as mediation and mapping. This paper presents a prototype method for understanding constraints in code. In particular, it focuses on understanding the constraints in EXPRESS information models, for which an experimental computerised system has been implemented. A casebased reasoning approach is taken, in which cases suggest so-called higher-level constraints to extract from matching fragments of EXPRESS code.