【摘要】 一篇论文往往具有严谨明确的脉络结构,而关键词又与论文主题息息相关。从这两个角度出发,面向论文领域,对传统的TextRank自动文摘算法进行改进,将关键词覆盖率、标题契合度和关键句位置等信息引入到自动文摘提取的算法之中。最后,通过两组对比实验,证明改进之后的算法比传统的TextRank算法具有更高的准确率和更低的召回率,同时具有更强的稳定性。更多还原

【Abstract】 Clear structure is a typical feature of paper, and keywords are closely geared to the paper’s idea. From these two aspects, proposes an improved method on traditional TextRank for automatic summarization, keywords coverage, similarity with title and the location of sentences were taken into account. Two contrast experiments prove that the improved method has higher accuracy rate and recall rate compared with traditional TextRank, besides, it is more stable.更多还原