Abstract: To extract the abstract with less redundant information and a wide coverage,which can reflect the main idea of the text,this paper advanced a comprehensive text summarization method.This method takes the frequency of the words,the title,the position of the sentence in the text,cue phrases,similarity of the sentences and other features in the text into consideration,constructs a comprehensive feature weighting function,trains the corpus with mathematical regression model,removes the redundant information,and then gets the abstract.The experiment shows that this method is very effective and feasible,and very superior in the quality of the extraction.

Key words: Automatic abstract,Features,Comprehensive,Weighting function