基于依存适配度的知识自动获取词义消歧方法  

Word Sense Disambiguation Based on Dependency Fitness with Automatic Knowledge Acquisition

作　　者：鹿文鹏[1,2] 黄河燕[1] 

机构地区：[1]北京理工大学计算机学院北京市海量语言信息处理与云计算应用工程技术研究中心,北京100081 [2]山东轻工业学院理学院,山东济南250353

出　　处：《软件学报》2013年第10期2300-2311,共12页Journal of Software

基　　金：国家自然科学基金(61132009);国家重点基础研究发展计划(973)(2013CB329303)

摘　　要：针对困扰词义消歧技术发展的知识匮乏问题,提出一种基于依存适配度的知识自动获取词义消歧方法.该方法充分利用依存句法分析技术的优势,首先对大规模语料进行依存句法分析,统计其中的依存元组信息构建依存知识库;然后对歧义词所在的句子进行依存句法分析,获得歧义词的依存约束集合;并根据WordNet获得歧义词各个词义的各类词义代表词;最后,根据依存知识库,综合考虑词义代表词在依存约束集合中的依存适配度,选择正确的词义.该方法在SemEval 2007的Task#7粗粒度词义消歧任务上取得了74.53%的消歧正确率;在不使用任何人工标注语料的无监督和基于知识库的同类方法中,取得了最佳的消歧效果.A word sense disambiguation （WSD） method based on dependency fitness is proposed to solve the problem of knowledge acquisition bottleneck in the development of WSD techniques. The method achieves automatic knowledge acquisition in WSD by taking full advantage of dependency parsing. First, a large-scale corpus is parsed to obtain dependency cells whose statistics information is utilized to build a dependency knowledge base （DKB）; then, the ambiguous sentence is parsed to obtain the dependency constraint set （DCS） of ambiguous words. For each sense of ambiguous word, sense representative words （SRW） are obtained through WordNet. Finally, based on DKB, dependency fitness of all kinds of SRW on DCS is computed to judge the right sense. Evaluation is performed on coarse-grained English all-words task dataset of SemEval 2007. Compared with unsupervised and knowledge-based methods which don＇t utilize any sense-annotated corpus, the proposed method yields state-of-the-art performance with F1-measure of 74.53%.

关 键 词：词义消歧 依存句法分析 知识获取 依存适配度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]