【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.更多还原

【Abstract】 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.更多还原