Inexact product structure matching algorithm based on multiplex domain semantics

Similarity of product structures is an inexact matching problem among complex information objects with inductive structures. The similarities related not only to deductive nodes, but also to attribute set associated with each node, and specific background product concept (ontology) system belonging to enterprises. The product concept system offered a rich set of background semantics knowledge to estimate the similarity. A triple characteristics model was constructed which was composed of attribute sets of product structure, product inductive structure and product background concept system. Based on this model, a comprehensive similarity computation model based on multiplex domain semantics was constructed, which was the combination of several sub-indicators, such as semantic distance, similarity among feature sets and similarity among inductive structures. Analysis showed that this model was a better choice to estimate the similarity among product structures. A non-recursion algorithm to realize the model was proposed, and the operation performance of the algorithm was studied, as well as the preconditions required for the algorithm to be applied in real world were also discussed.