Adaptive Chinese Combinatorial Ambiguities Disambiguate Method

Combinatorial ambiguity has always been a vital issue in Chinese word segmentation. This paper presented a novel way for disambiguation by use of a multi maximal log likelihood ratio of the cooperative statistical table, which took the cooperative examples provided by the artificial checked word segmentation as the initial cooperative knowledge of covering ambiguity. Then the key factors were determined based on the experiments regarding combinatorial ambiguities (the size of the context windows, the sensitivity of locations in the windows as well as weighting of feature word, et al). On this basis, an adaptive Chinese combinatorial ambiguities disambiguate method was used to strengthen and stabilize collocations. At lastly, the method was tested and proved that it can well raise accuracy.