基于HowNet的术语语义知识库构建技术  

Construction techniques of terminology semantic knowledge base based on HowNet

作　　者：王羊羊[1] 陈刚[2] 蔡东风[1] 王裴岩[1] 

机构地区：[1]沈阳航空航天大学人机智能研究中心,沈阳110136 [2]中航昌河飞机工业(集团)有限责任公司科技部,江西景德镇333000

出　　处：《沈阳航空航天大学学报》2016年第4期78-84,共7页Journal of Shenyang Aerospace University

基　　金：国防基础科研项目(项目编号:A0520131003);辽宁省高校创新团队支持计划(项目编号:LT2014005)

摘　　要：领域知识库能够满足特定领域的自然语言处理系统对知识的需求,然而大部分领域知识库的构建方式为手工构建,效率较低。针对这一问题,分析已经手工构建的2 300余条航空术语描述信息及其在构建过程中总结的规则,在此基础上,总结了200余条核心词框架,核心词以外的其他词,通过一种规则与统计相结合的方法进行框架的自动填充,从而提高了构建术语语义知识库的自动化程度。最后,对构建的术语描述信息进行了相似度计算,取得了较好的结果。Knowledge base for specific domains can satisfy the knowledge requirements for the natural language processing system. However,most current work for building the domain knowledge base is hand-built and inefficient. To solve this problem,more than 2300 pieces of terminology describing information and the rules that have been formulated were analyzed. Then more than 200 semantic frameworks were summarized based on core word while the other words are filled to the frameworks according to the rule and the statistical results. Finally,the validness of the constructed knowledge base is demonstrated that good results are achieved in term of similarity calculation.

关 键 词：术语 规则 统计 语义知识库 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]