Improving Chinese Tokenization With Linguistic Filters On Statistical Lexical Acquisition