Studies on automatic recognition of Chinese adverb CAI's usages based on statistics

Studies about the Functional Words Knowledge Base began in recent years. It has gotten some achievements. The functional words include adverbs, preposition, conjunction, auxiliary, and modality. The “Trinity” knowledge-base of functional words has been initially built which includes function usage dictionary, usage rules-base and usage corpora. This paper bases on the previous work, and further study automatically recognizing Chinese adverb CAI's usages using statistical methods. Two statistical models, viz. CRF and ME, are used to tag the adverb CAI's usages on the tagged corpus of People's Daily(1998.1). The precision rate of CRF and ME in opening test is 73.8% and 73.9% respectively. In closing test the precision rate of both are 100%. The experiments show that statistic-based method is more effective in usage automatic recognition of the adverb CAI than the rule-based method.