关键词: 主题句抽取，无向图，文档句权重，自动文摘

Abstract: Undirected graph based on the sentence was proposed. The problem of sentence extraction was transformed to computing undirected graph node weights. This paper first proposed sliding window-based keywords extraction algorithm,followed by the establishment of the undirected graph. The edge weights of the graph were modeled by the Vecfor Space Model(VSM) in turn. The node weights were computed finally by the weight model based on the similarity matrix,and the subject sentences were obtained on the ratio of compression. Experiments show that the proposed automatic summarization techniques improve the recall rate and accuracy effectively.

Key words: Subject sentence extraction, Undirected graph, Sentence weight, Automatic text summarisation