Graph-Based Chinese Word Sense Disambiguation with Multi-Knowledge Integration

作     者：Wenpeng Lu Fanqing Meng Shoujin Wang Guoqiang Zhang Xu Zhang Antai Ouyang Xiaodong Zhang 

作者机构：School of Computer Science and TechnologyQilu University of Technology(Shandong Academy of Sciences)Jinan250353China School of Information Science and EngineeringZaozhuang UniversityZaozhuang277160China Department of ComputingMacquarie UniversitySydneyNSW 2109Australia Centre for AudioAcoustics and VibrationUniversity of Technology SydneySydneyNSW 2006Australia School of Computer Science and EngineeringShandong University of Science and TechnologyQingdao266590China Jinan Intellectual Property Information CenterJinan250099China 

出 版 物：《Computers, Materials & Continua》 (计算机、材料和连续体（英文）)

年 卷 期：2019年第61卷第7期

页      面：197-212页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：The research work is supported by National Key R&D Program of China under Grant No.2018YFC0831704 National Nature Science Foundation of China under Grant No.61502259 Natural Science Foundation of Shandong Province under Grant No.ZR2017MF056 Taishan Scholar Program of Shandong Province in China(Directed by Prof.Yinglong Wang) 

主　　题：Word sense disambiguation graph model multi-knowledge integration word similarity 

摘      要：Word sense disambiguation(WSD)is a fundamental but significant task in natural language processing,which directly affects the performance of upper ***,WSD is very challenging due to the problem of knowledge bottleneck,i.e.,it is hard to acquire abundant disambiguation knowledge,especially in *** solve this problem,this paper proposes a graph-based Chinese WSD method with multi-knowledge ***,a graph model combining various Chinese and English knowledge resources by word sense mapping is ***,the content words in a Chinese ambiguous sentence are extracted and mapped to English words with ***,English word similarity is computed based on English word embeddings and knowledge *** word similarity is evaluated with Chinese word embedding and HowNet,*** weights of the three kinds of word similarity are optimized with simulated annealing algorithm so as to obtain their overall similarities,which are utilized to construct a disambiguation *** graph scoring algorithm evaluates the importance of each word sense node and judge the right senses of the ambiguous *** experimental results on SemEval dataset show that our proposed WSD method significantly outperforms the baselines.