After the machine recognition of chinese speech or in the input process of Chinese characters by Using the Chinese phonetic alphabets (PINYIN), the difficulty of determining the Chinese characters denoted by each PINYIN unit is caused by the homonymous nature of Chinese characters. The present paper discusses method for converting the Chinese phrases in the form of phonetic sequences to phrases represented by Chinese characters based on syntac tic analysis to verify their correct segmentation. For tfils purpose, specifically developed ATNs (Augmented Transition Networks) are applied to their syntactic parsing. The problems caused by numerous uncertainties in both phrase segmentation and homonym distinguishing are solved by means of-modifying the construction of ATNs and characterizing the features of Chinese language in the process of matching.
展开▼