基于语义依存分析的CFN框架排歧  

CFN frame disambiguation based on semantic dependency analysis

作　　者：门宇鹏 郝晓燕 董嘉敏 MEN Yu-peng;HAO Xiao-yan;DONG Jia-min(College of Information and Computer,Taiyuan University of Technology,Taiyuan 030600,China)

机构地区：[1]太原理工大学信息与计算机学院

出　　处：《计算机工程与设计》2019年第9期2654-2659,共6页Computer Engineering and Design

基　　金：教育部人文社会科学研究基金项目(17YJA740031);国家自然科学基金项目(61503273);山西省自然科学基金项目(201801D121137、201701D22111252)

摘　　要：为大规模自动构建语料库,使计算机能够理解文本信息,提出框架自动识别,框架排歧作为框架识别的子任务,是亟待解决的。框架排歧即根据目标词的上下文信息,从现有的框架库中,自动为该目标词标注一个合适的框架。利用分类算法的思想,使用支持向量机(SVM)中的“一对一”方法建立多个分类器来解决多分类问题。通过向已有的词性与依存句法关系特征之上再加入语义依存分析关系特征来提升框架排歧的准确率。训练语料和测试语料为山西大学建立的汉语框架语义知识库(CFN)中的7个词元,1519条例句,采用5-fold交叉验证。实验取得的最好结果为80.67%,验证了语义依存分析对框架排歧的有效性。To automatically construct a corpus on a large scale,and enable the computer to understand the text information,the automatically framework reorganization was presented.The framework disambiguation as a sub-task of the framework recognition is urgently needed to be solved.Using the idea of the classification algorithm,multiple classifiers were built using the one-to-one method in support vector machine(SVM)to solve the multi-classification problem.The accuracy of frame disambiguation was improved by adding semantic dependency analysis characteristics to existing features of word-of-speech and dependency syntax.The training corpus and test corpus were 7 vocabulary and 1519 example sentences in the Chinese FrameNet(CFN)established by Shanxi University,and 5-fold cross-validation was used.The best result of the experiment is 80.67%,which verifies the effectiveness of semantic dependence analysis on framework disambiguation.

关 键 词：汉语框架语义知识库(CFN) 框架排歧 多分类 支撑向量机 语义依存分析 交叉验证 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]