基于框架语义的隐式篇章关系推理  

Implicit Discourse Relation Inference Based on Frame Semantics

作　　者：严为绒 朱珊珊[1] 洪宇[1] 姚建民[1] 朱巧明[1] 

机构地区：[1]苏州大学计算机科学与技术学院,江苏苏州215006

出　　处：《中文信息学报》2015年第3期88-99,共12页Journal of Chinese Information Processing

基　　金：国家自然科学基金(61373097;1003152;61272259;60970056;60970057;90920004);教育部博士学科点专项基金(2009321110006;20103201110021);江苏省自然科学基金(BK2011282);江苏省高校自然科学基金重大项目(11KJA520003);苏州市自然科学基金(SYG201030;SH201212)

摘　　要：篇章关系分析是一种专门针对篇章语义关系及修辞结构进行分析与处理的自然语言理解任务。隐式篇章关系分析是其中重要的研究子任务,要求在显式关联线索缺失的情况下,自动检测特定论元对之间的语义关系类别。目前,隐式篇章关系分析性能较低,主流检测方法的准确率仅约为40%。造成这一现状的主要原因是:现有方法脱离论元的语义框架进行关系分析与检测,仅仅局限于特定论元特征的关联分析。针对这一问题,该文提出一种基于框架语义的隐式篇章关系推理方法,这一方法有效利用了框架语义知识库(即FrameNet)和相关识别技术,实现了论元语义框架的自动识别,并在此基础上,借助大规模文本数据中框架语义关联关系的分布概率,进行论元语义一级的关系判定。实验结果显示,仅仅利用第一层框架语义知识,即可提高隐式篇章关系检测性能至少5.14%;同时,在考虑关系类别平衡性的情况下,这一方法能提高至少10.68%。Discourse relation analysis is a task of natural language understanding which aimed at analyzing and dispo sing the semantic relation and rhetorical structure of discourse. Implicit discourse relation analysis is an important subtask of automatically detectind senses of semantic relation between arguments in the absence of direct cues. Cur- rently, the performance of implicit discourse relation analysis is low and state of-art accuracy can only reach 40%. The major cause of this situation is that the existing methods did not analyze arguments in the semantic frame, limit- ed only to the local features and correlation analysis of arguments. This paper proposes a method of implicit dis- course relation inference based on frame semantic. This method automatic recognised semantic frame of arguments through FrameNet and related identification technology. On this basis, we indentify the semantic relation of argu- ments by the distribution probability of frame semantic relation in large-scale text data. The experimental results show that, only using the first level of frame semantic can improve the detection performance of implicit discourse relation up to 5.14% ; meanwhile, this method can make the accuracy rate increased by 10. 68% in the case of con- sidering the balance of relation categories.

关 键 词：篇章关系 隐式篇章关系 框架语义 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]