Abstract: Sentence ranking is the key issue of text automatic summarization．Based on mutual reinforcement principles,we proposed a new sentence ranking model——paragraph-sentence mutual reinforcement model．With the relation between paragraphs and the mutual reinforcement between paragraphs and sentences,it iteratively computes the salience of the sentences and extract the summary sentences．We analyzed the effect of the internal and external reinforced factor and discussed the problem of redundancy remove．Experiments show that it can extract high quality summary when it applies to the single-document summarization.

Key words: Sentence ranking,Mutual reinforcement principle,Automatic summarization