Measuring product semantic similarity by exploiting a manufacturing process ontology

Abstract

The retrieval of manufacturing knowledge in companies is critical because product and process knowledge was not actually managed but only documented. Particularly, the identification of similarities between new and past products relied almost exclusively on the memory and the experience of people, and thus it is a time-consuming task. In this paper, a method to allow the automatic identification of past similar products is proposed, so that they can be used to speed up the design of manufacturing of the new product. The similarity is computed by using a semantic model in the form of ontology, which constitutes the hierarchical structure of concepts. A new similarity index is defined based on the portion of overlapping subgraph of concepts existing between two products. The different weight of each node is also considered because more descendants the node has, less specific its semantics information content is. The computation of the similarity measurement will allow the discovering of knowledge from stored data, thus supporting the engineers in searching for past products having similar characteristics with the new one. The potentiality of the proposed index is shown in a motivating example.