首页 | 本学科首页   官方微博 | 高级检索  
     

多知识源融合的自动摘要系统研究与实现
引用本文:陈燕敏,王晓龙,刘秉权,楼喜中. 多知识源融合的自动摘要系统研究与实现[J]. 高技术通讯, 2006, 16(4): 337-341
作者姓名:陈燕敏  王晓龙  刘秉权  楼喜中
作者单位:哈尔滨工业大学计算机科学与技术学院,哈尔滨,150001;哈尔滨工业大学计算机科学与技术学院,哈尔滨,150001;哈尔滨工业大学计算机科学与技术学院,哈尔滨,150001;哈尔滨工业大学计算机科学与技术学院,哈尔滨,150001
基金项目:中国科学院资助项目 , 国家科技攻关项目
摘    要:提出并实现了一种能有效地融合改进的浅层主题特征分析方法、词汇链方法、话语结构方法的分析结果来生成文摘的多知识源融合的自动摘要系统,并对其进行了评测实验.评测结果表明,该系统在捕获文章特征的同时较好地保持了原文的内容及内在逻辑结构,生成的文摘具有良好的连贯性和流畅性;与采用单一方法的自动摘要系统相比较,生成的文摘质量有明显提高.

关 键 词:自动摘要  特征词  词汇链  话语结构  融合
收稿时间:2005-04-18
修稿时间:2005-04-18

Research and design of summarization system based on multi-knowledge sources fusion
Chen Yanmin,Wang Xiaolong,Liu Bingquan,Lou Xizhong. Research and design of summarization system based on multi-knowledge sources fusion[J]. High Technology Letters, 2006, 16(4): 337-341
Authors:Chen Yanmin  Wang Xiaolong  Liu Bingquan  Lou Xizhong
Affiliation:School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001
Abstract:An automatic text summarization system based on fusing multiple knowledge sources is proposed. It generates summaries by fusing the results produced by the improved method based on add terms, the improved method based on lexical chains and the improved method based on discourse structure. Evaluation results show that the presented system not only captures the characteristics of texts but also retains the content and inherent logic of sources, and high-quality summaries are produced from arbitrary Chinese texts. The proposed system is compared to systems based on single method and it is shown that it produces either comparable or better summaries overall.
Keywords:automatic text summarization   add term   lexical chain   discourse structure   fusion
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号