专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明专利

AU2021101672A4 FUZZY LOGIC-BASED QUERY EXPANSION METHOD FOR EFFICIENT INFORMATION RETRIEVAL USING RELEVANCE FEEDBACK APPROACH 未知
公开(公告)号：AU2021101672A4
公开(公告)日：2021-05-20
申请号：AU2021101672
申请日：2021-03-31
申请人： AGRAWAL GAURAV , KUMAR VINOD , MAHESHWARI SAUMIL , SAINI MAYANK , SAJID MOHAMMAD , SHARMA BRIJESH CHAND , SINGH JAGENDRA
发明人： SINGH JAGENDRA , SHARMA BRIJESH CHAND , SAJID MOHAMMAD , SAINI MAYANK , KUMAR VINOD , MAHESHWARI SAUMIL , AGRAWAL GAURAV
IPC分类号： G06F16/33 , G06N5/04
摘要： The present disclosure relates to a method for QE based on fuzzy logic considering top-retrieved document as relevance feedback documents for mining additional QE terms. The method includes mining additional QE terms upon considering top-retrieved document as relevance feedback documents by employing QE based on fuzzy logic; mining additional expansion terms upon calculating the degrees of importance of all unique terms of top-retrieved documents collection by selecting different QE terms selection methods; obtaining different relevance scores for each term; combining different weights of each term by using fuzzy rules to infer the weights of the additional query terms; and forming a query vector using the weights of the additional query terms and the weights of the original query terms to retrieve documents. 100> retrievirg relevant documents for original user query ina first iteration by using BM25 similarityfuriction with empirical pararnetersettirig 1 10 selectingtop n retrieved documents as PRF documents selectingall unique terms ofIPRFdocuments as thecandidate term setor term pool applyingstatistical-based, class-basedaridco-occurrence-based weightingapproaches for providing scores toall candidateterms cornbiringcandidate terms weightsobtained from statistical, class and co-occurrence-based approaches using three 108 fuzzylogiccontrollers (FLCs)infirstlevel fuzzy inferencesystem (FIS) combinirgweightingscores of FLC F arid _ usingone FLC in second level FISandcorbining W W andW weightingscores usingFLCj com biring weighting score of TFIDF with the weight obtained frornFLC, usingone LCin third-level FIS filteringnorn-relevant candidate terms upon applyrngsemanticfilteringapproach, whereinsemanticfilteredcandidate termisselected as additional query termsand used for expanding the userquery reweighting additional expansion terms by employing reweighting module "118 retrievirg ra nked list of releva nt documents for expa ended user query using additiona Iquery termssggested by FLBQEarid FLSBQE techniques Figure1 - ------------- r ------------------------- I Userquery arngFLCSMastUc Statuibased~ehd Search k .. CHI andu10 FLc eas FLCcom d C sgubased2Meth 1ACCAD andSecond eved FIS FMal retrieved we ranked docs Coo~ccurrence MetOdais: : asm 1.vel Fns Term PannF Methods Fuzzy Inration System Quay exp"en"de' FI.Cfmal roanultion queing t*ems TROF expanuien Fuzzy Infonnaton System Figure 2

2. 发明专利

AU2021102702A4 A PROCESS FOR QUERY REFORMULATION SYSTEM USING RANK AGGREGATION AND GENETIC APPROACH 未知
公开(公告)号：AU2021102702A4
公开(公告)日：2021-11-18
申请号：AU2021102702
申请日：2021-05-20
申请人： KUMAR RAKESH MR , KUMAR SAURABH MR , SINGH DEEPAK DR , SINGH JAGENDRA DR , YADAV ARUNA MS
发明人： SINGH JAGENDRA
IPC分类号： G06F16/332 , G06F16/33
摘要： The present disclosure relates to a process for query reformulation system using rank aggregation and genetic approach. In the present disclosure the power of combining multiple query expansion terms selection methods is explored to improve the performance of information retrieval system by using the automatic expansion of user query called automatic query expansion (AQE). The ranks combination of four query expansion terms selection methods, which are Chi-Square Statistic (CHI), Co-occurrence Information (Co-occurrence), Binary Independence Model (BIM) and Robertson Selection Value (RSV) on two real datasets with or without semantic filtering and semantic genetic filtering approaches. The results with real data sets demonstrates that combining multiple QE terms selection methods could improve the performance of AQE in terms of Mean Average Precision, Recall, and F-Measure. performing pseudo-relevancefeedback based automatic query expansion us ing CHI, Co- /,Av102 occurrence, B IM a nd RSV terms selection methods bI combining ranked list of query expansion terms suggested by different expansion terms selection 104 methods using rank agg regation methods filtering out semantically irrelevantand redundant expansion terms with context to user query 106 obtained after com boning multiple termsselecon methods throughsemantifiltering approach Generating anoptimal combination of query terms by geneticapproach and obtaining candidate 1A i expansionterms by applying rankscombination andsemanticfiltering approach. Figure1 Figure 2

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式