Automatic comparison of design knowledge generated and consumed during a design process is very useful. Basically, design knowledge can be classified into four categories: (a) artefact (product) structures, (b) artefact behaviours, (c) artefact functions and (d) causalities among structures, behaviours and functions. This paper proposes an intelligent hybrid graph approach to represent design knowledge associated with these categories. Comparison of design knowledge thus turns out to be the comparison of hybrid graphs. A general algorithm is developed for the graph comparison purpose. In this general algorithm, comparison of line graphs is an important task, which is realized by using the Hopfield—Tank neural network algorithm. An initial implementation of this algorithm proves to be promising.
Graph comparison via the nonbacktracking spectrum
