Abstract: We propose a word segment algorithm based on word group. In this model, word group is used for Ambiguity analysis. At first step, statistical information is used for build information base. In the process of dealing with sentence, a small step is triggered for counting information of adjoining situation and word frequency and calculates parameters of this model according to size of window. When get different word sequences, we use Analysis Tree to find the prime sequence. Because of short in decision distance, we get a low time complexity. Algorithm analyzing and result of experiment show that segmentation algorithm based on word group has higher efficiency and accuracy.