引入IVHFS的数学表达式检索结果排序方法  

Ranking Method of Mathematical Expression Retrieval Results with Interval Valued Hesitation Fuzzy Sets

作　　者：韩苹 田学东 HAN Ping;TIAN Xue-dong(School of Cyber Security and Computer,Hebei University,Baoding 071002,China)

机构地区：[1]河北大学网络空间安全与计算机学院,河北保定071002

出　　处：《小型微型计算机系统》2021年第1期34-40,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61375075)资助;河北省自然科学基金项目(F2019201329)资助;河北省教育厅河北省高等学校科学技术研究重点项目(ZD2017208)资助。

摘　　要：针对数学表达式表达形式多样、语法语义变换丰富给基于数学表达式的检索结果排序所带来的困难,提出一种基于IVHFS(Interval Valued Hesitation Fuzzy Sets)的数学表达式检索结果排序算法,利用IVHFS在排序中可以完整保留属性信息的优势,改善排序性能.首先,通过对数学表达式检索特征的归纳,确定数学表达式检索结果的排序属性;然后,利用所归纳的数学排序属性构造IVHFS集合;最后,采用IVHFS相似性测度计算出数学表达式之间的相似度,进而得出排序结果.本实验从公共数据集NTCIR-12_Math IR_Wikipedia_Corpus中获取了528188个数学表达式作为实验数据集,并设计了子式空间结构属性、运算符关联属性以及运算数关联属性作为评价指标,从而实现数学表达式检索结果的排序.实验结果表明,检索系统的查全率和查准率分别为75.8%和66.4%,其检索结果排序效果更加合理.To solve the ranking problems of mathematical expression retrieval hits resulted from the various styles,rich grammatical and semantic transformation of mathematical expressions,a ranking algorithm of mathematical expression retrieval results based on IVHFS(Interval Valued Hesitation Fuzzy Sets)is proposed,w hich can make use of the advantages of IVHFS to retain attribute information completely in sorting and improve sorting performance.First,based on the induction of the characteristics of mathematical expressions,the sorting attributes of the retrieved results are determined.Then,construct the IVHFS sets by using the summarized the sorting of math attributes.Finally,use the IVHFS similarity measures to calculate the similarity betw een mathematical expressions and then get the sort of results.In order to achieve the sorting of mathematical expression retrieval results,528188 mathematical expressions w ere obtained from the public dataset NTCIR-12_M ath IR_Wikipedia_Corpus and subspace structure attribute,operator associated attribute and operand associated attribute are designed as the evaluation indicators in this experiment.Experimental results show that the recall and precision of the method are 75.8%and 66.4%,and the sorting effect of retrieval results are more reasonable.

关 键 词：数学检索 检索结果排序 区间值犹豫模糊集 多属性决策 相似度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]