一种基于电阻网络的文本摘要方法  

A text summarization method based on resistance network

作　　者：叶静[1] 贺明科[1] 谭文堂[1] 吴昊[2] 李芳芳[1] 

机构地区：[1]国防科技大学信息系统与管理学院,湖南长沙410073 [2]北京系统工程研究所北京100101

出　　处：《中国电子商情（通信市场）》2012年第2期103-108,共6页

摘　　要：句子打分算法作为抽取式自动摘要方法中最为关键的组成部分，成为近年来自动摘要领域的热点问题。本文提出了一种基于电阻网络图的摘要句子打分方法，该方法将句子作为节点，句子之间的相似度作为边，通过基尔霍夫定律在文本相似度图上计算两节点间的电阻距离，并在此基础上提出一种基于电能消耗的节点重要度评估方法来抽取文章句子形成摘要。本文还给出了一个基于该模型的高效算法，最后通过实验分析验证了通过该模型和算法来生成文本摘要的高效性和有效性。the sentence scoring algorltt^m as the most crucml part m automatic extraction summarization methods, has been the hotspot in the field of automatic Summarization in recent years. This paper puts forward a sentences scoring methods based on the resistance of the network diagram, this method regard a sentence as a node, the similarity between sentences as edge, then calculating resistance distance of two nodes in text similarity graph through the kirchhoffs law, and based on this, puts forward a kind of based on electricity consumption of node importance evaluation method to extract the sentence to form summarization. This paper also gives a efficient algorithm based on this model, the experimental analysis verified the efficiency and effectiveness of the model and algorithm.

关 键 词：电阻网络 摘要 句子打分 节点重要性 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]