Mapping Items between Information Classification Standards Based on Semantic and Structure Similarities: a Pilot Study

Automatic mapping items between different versions of classification standards is challenging and has great values in practice, as these information classification standards play key roles in data exchange between different application systems in e-government, e-commerce, and other domains. In this paper, we propose an automatic method for mapping items between standards based on semantic and structure similarities: starting with mapping items of the same name, and employing thoroughly the semantic constraints embedded in the hierarchical category structures. Experiments on the Chinese national standard of classification and codes for fixed assets demonstrate that the proposed method could help to finish more than 74% of the task for mapping items between two versions of that Chinese national standard (v1994 and v2010), and is expected to dramatically reduce the time and lower the cost of manual mapping by domain experts.