Chinese person name recognition with semantic radical

Radical is the primary component of Chinese character, but its function in Chinese information processing has not yet been well studied. With both theoretical analysis and corpus statistics, this paper presents the correlation between semantic radicals and Chinese characters used in person names. A person name recognition method that uses semantic radicals as computational resources in maximum entropy framework is proposed. This method was contrasted with a baseline method that doesn´t take radicals into consideration. The experiments results show that semantic radicals can be quite useful in Chinese text processing.