基于模糊决策的主旨句生成算法

提炼文档的主旨句在自动文本摘要生成系统中处于十分重要的地位.传统方法给出一个计算句子权重的公式,但此公式形式单一固定,不够灵活,通过借鉴模糊决策理论,提出了一种基于模糊决策的主旨句生成方法.该方法不局限于影响句子权重因素的数量及形式,它依据模糊决策理论对文档中每个句子的重要程度做出综合评判,将那些较大程度上反映文章主旨的语句提炼出来,作为摘要的生成语句.由于人们在主观上对各个因素的重视程度也有所不同,因此,又提出一种改进算法.实验结果表明,给出方法所提取的文章主旨语句较传统公式法的准确性大大提高了.

Algorithm of producing topic sentence based on fuzzy decision

Topic sentence abstraction plays an important role in automated text summarization system.A conventional method is a fixed and inflexible formula that computes the importance of each sentence.In order to overcome the shortcoming of the formula,a novel method for judging and computing the importance of sentences based on fuzzy decision theory is presented.Sentences are selected in ac-cordance with their relative importance and organized in a readable way as the topic sentences to form the final abstract.In fact,the factors of affecting the importance of sentences are complicated.The number of the factor is alterable and the degree of attaching impor-tance to each factor is different.Hence,an improving algorithm is also given to solve such problem.Experiments show that the sentences are selected more accurately.