RDF问答系统中一种基于N-gram的消歧方法  

Approach of Disambiguation Based on N-gram in RDF Question Answer System

作　　者：江伟豪 严丽 屠要峰[2] 周祥生[2] 李忠良 JIANG Wei-hao;YAN Li;TU Yao-feng;ZHOU Xiang-sheng;LI Zhong-liang(College of Computer Science and Technology/College of Artificial Intelligence,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China;ZTE Corporation,Nanjing 210000,China)

机构地区：[1]南京航空航天大学计算机科学与技术学院/人工智能学院,南京211106 [2]中兴通讯股份有限公司,南京210000

出　　处：《小型微型计算机系统》2022年第5期969-975,共7页Journal of Chinese Computer Systems

基　　金：江苏省基础研究计划项目(BK20191274)资助。

摘　　要：由于知识网络与互联网应用的高速发展,RDF(Resource Description Framework,资源描述框架)被广泛应用到关联数据的存储以及知识图谱的创建当中.基于自然语言处理的RDF问答系统是普通用户查询RDF数据的高效方法.在处理自然语言的过程中一般分为用户意图理解和查询验证两个阶段.而现存的研究方法是在用户意图理解阶段使用联合消歧的方式消除歧义,并且在查询验证阶段进行穷举验证,无效语句的运行延长了响应时间.本文基于N-gram模型建立语义概率模型,利用语义概率模型在用户意图理解阶段解决结构歧义与映射歧义的问题,且最终将查询意图转化为top-k个最优的查询语句进行查询并获取结果.通过与现存的方法在真实基准数据集中测试对比,本方法提高了在解决隐式关系问题方面的准确率,并且提升了查询性能.With the rapid development of knowledge network and Web applications,RDF(Resource Description Framework)is widely used in the storage of linked data and the creation of knowledge graphs.RDF question answering system with natural language processing is an efficient way for ordinary users to query RDF data.In general,the process of natural language processing is divided into two stages:user intention understanding and query validation.The existing research methods generally use joint disambiguation in the user intent understanding phase and exhaustive validation in the query validation phase,where invalid statements are run to prolong the response time.In this paper,a semantic probability model is established based on the N-gram language model,and the semantic probability model is used to solve the problem of structural ambiguity and mapping ambiguity at the stage of user intention understanding.Finally,the query intent is transformed into top-k query statements to obtain the results.By comparing the existing methods tested in real benchmark datasets,our approach improves the accuracy of solving implicit relational problems and speeds up the query performance.

关 键 词：RDF问答系统 N-GRAM语言模型 自然语言处理 消歧 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]