节点文献
一种面向研讨环境的摘要生成方法
Probabilistic Mixture Model Based Summarization Approach for CWME Discussions
【摘要】 针对当前研讨厅中对专家大段发言缺乏归纳、概括方法的问题,提出了一种面向研讨环境的摘要生成方法,该方法采用概率混合模型抽取专家发言的话题集,对相邻话题的变化情况进行判断,不仅可以发现话题(Topic)的转变趋势,还可根据话题的演化生成相应的摘要,提供给专家。这些自动生成的摘要既有助于增强专家之间的良性互动、激发专家思维,也可用于决策方案和会议总结的辅助生成。实验结果表明,提出的面向研讨环境的摘要生成方法合理、有效。
【Abstract】 To solve the problem of the lack of summarization method of experts’ speeches in CWME,a new method was proposed.This method first adopts a probabilistic mixture model to discover latent themes from experts’ speeches.Then,it judges whether themes have changed.Once they are changed,this method will generate summary and provide it for experts.The summaries can help experts to inspire each other in CWME.Experimental result shows that the proposed method is feasible and effective.
【关键词】 综合集成研讨厅;
概率混合模型;
群体交互;
摘要生成;
【Key words】 Cyberspace for workshop of metasynthetic engineering(CWME); Probabilistic mixture model; Collective interaction; Text summarization;
【Key words】 Cyberspace for workshop of metasynthetic engineering(CWME); Probabilistic mixture model; Collective interaction; Text summarization;
【基金】 973国家重点基础研究发展计划(2007CB311007)资助。
- 【文献出处】 计算机科学 ,Computer Science , 编辑部邮箱 ,2011年02期
- 【分类号】N941.4
- 【被引频次】3
- 【下载频次】73