Using directed graph based BDMM algorithm for Chinese word segmentation

Abstract

Word segmentation is a key problem for Chinese text analysis. In this paper, with the consideration of both word-coverage rate and sentence-coverage rate, based on the classic Bi-Directed Maximum Match (BDMM) segmentation method, a character Directed Graph with ambiguity mark is designed for searching multiple possible segmentation sequences. This method is compared with the classic Maximum Match algorithm and Omni-segmentation algorithm. The experiment result shows that Directed Graph based BDMM algorithm can achieve higher coverage rate and lower complexity. © Springer-Verlag Berlin Heidelberg 2005.