A New Parallel Segmentation Model Based on Dictionary and Mutual Information

It is difficult to compute the word frequency for mutual information segmentation. Statistic of word frequency of parallel mutual information is integrated with dictionary segmentation to improve efficiency in this paper. The parallel model and dispatching policy are presented, the paper also gives the speed up ratio of parallel model at the same time, periods pattern string and non periods pattern string are optimized in parallel model. Experiment show that the algorithm is available. The parallel model also can use for other segmentation algorithms that base on statistic of word frequency.