An Automatic Approach for Domain-Specific Dictionary Expansion Based on Web Mining

This paper proposes an automatic expansion approach for an existing domain-specific dictionary based on Web mining. Using the terminology pairs in a dictionary as queries, we first extract snippet fragments that potentially contain the Chinese translations of current English phrases. Based on matching patterns extracted from the Web, we can get the most likely translation for a new English phrase. Finally we use the vector space model to filter out the translation equivalents belong to our expected domain, and a domain-specific dictionary expanded by these new terminology pairs is thereby built. The performance of our approach is verified on a dictionary of finance and accounting, and a precision between 85-90% is achieved on considering different thresholds.