An approach of Multi-document Summarization Based on Text Relationship Map

摘要：

This paper presents a novel extractive approach based on text relationship map to multi-summarization task. The whole system can be divided into three parts: TextTiling used to partition each document into segments; vector space based on concept segment similarity computation; and finally ranking score is obtained for each sentence after the construction of Text Relationship Map. Experiments on DUC are performed about the ROUGE evaluation results demonstrate that the proposed approach is an effective way for multi-document summarization. Experiments results also show that using HowNet for automatic word sense disambiguation has greatly improved the accuracy of the sentence similarity computation.

关键词： Multi-document summarization TextTiling HowNet Text Relationship Map

作者: Huifang Ma Weizhong Zhao Zhongzhi Shi

作者单位: Key laboratory of Intelligent Information Processing,Institute of Computing Technology,Chinese Academy of Science No.6 Kexueyuan Nanlu.10080 Beijing,China

会议类型: 国际会议

会议名称: 2008高等智能国际会议(2008 International Conference on Advanced Intelligence)

会议地点: 北京

会议语种:英文

在线出版日期: 2008-10-18（万方平台首次上网日期，不代表论文的发表时间）