A Model for Chinese Sentence Ordering Based on Markov Model

In this paper, we discuss a method to improve the sentence ordering task in Chinese. The way we approach is based on the analysis of Markov model, which can train transition probability in raw corpus. We iteratively calculate the largest transition probability path in Markov model to confirm the correct order. The method avoids judging the first sentence, which could lead to an instable result in our early work. We also provide a way to evaluate the effect of experiments. Experimental results indicate that our method shows good results on accuracy, and significantly improves the readability and coherence of the article. The method could be used in various fields of Chinese text processing work and applications.