Full-Text Search:
Home|Journal Papers|About CNKI|User Service|FAQ|Contact Us|中文
《China Sciencepaper》 2017-14
Add to Favorite Get Latest Update

Rule-based algorithm for sentence similarity computing

ZHANG Zixian;ZHANG Dedong;School of Mechanical,Electronic & Information Engineering,China University of Mining & Technology;School of Foreign Language Education,Liaocheng University;Beijing Feishun Information Technology Co.,Ltd.;  
A rule-based similarity computing method for sentence is proposed in this paper.In this method,user question(defined as Q1)is compared with all questions(defined as Q2)in the question data base and the answer of the question with highest similarity is given to the user.The five aspects include key word similarity,inverse order,key word distance,the ratio of key words in Q1 to those in Q2 and the ratio of functional words in Q1 to those in Q2.In experiment,this method is applied to online talk robot Xialiao(FreeTalker),and 65 534 question-answer pairs are taken from data base.Then similarity between Q1 and Q2 is computed.It is found that 39.75% cases of similarity fall within the range(0.6,0.7),which makes the robot relatively efficient in communication with people.
【Fund】: 聊城大学外语教育学院科研项目
【CateGory Index】: TP391.1
Download(CAJ format) Download(PDF format)
CAJViewer7.0 supports all the CNKI file formats; AdobeReader only supports the PDF format.
©2006 Tsinghua Tongfang Knowledge Network Technology Co., Ltd.(Beijing)(TTKN) All rights reserved