Auto-acquisition method for fine-grained semantic relations of commodity

To solve the problem of coarse-grained ontology model and lack of fine-grained semantic relations for commodity in application of electronic commerce, this paper proposes an idea of extracting classification feature from the vocabularies of product´s candidate properties, and an automatic acquisition method for fine-grained semantic relations based on supervised learning. According to the practical data, the correct classification rate of commodity reaches 86.05%, its average accuracy also reaches 83.9%, which turns out the effectiveness and feasibility of the proposed approach.