A new matching algorithm for Chinese place names

Abstract

Matching algorithm for Place Names is one of the most important research topics in the construction of digital gazetteers. Taking account of the morphological characteristics of Chinese Place Names and ontology method, we first pretreat the Chinese Place Names to remove illegal characters, decompose the Chinese Place Names into special names and generic terms, and then use both the Levenshtein Distance(LD) method and the semantic distance method to calculate relationship of similarity between them. Finally, we get the comprehensive similarity for Chinese Place Names by calculating the weighted average for both similarity of special names and that of generic terms. In this study, the example shows that similarity index for special names, generic terms and the overall place names, enhance the completeness and accuracy of the place names matching theory.