Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/108315
PIRA download icon_1.1View/Download Full Text
DC FieldValueLanguage
dc.contributorDepartment of Computingen_US
dc.creatorBian, Zen_US
dc.creatorYan, Xen_US
dc.creatorZhang, Jen_US
dc.creatorYiu, MLen_US
dc.creatorTang, Ben_US
dc.date.accessioned2024-08-02T08:47:20Z-
dc.date.available2024-08-02T08:47:20Z-
dc.identifier.isbn979-8-3503-1715-2en_US
dc.identifier.urihttp://hdl.handle.net/10397/108315-
dc.language.isoenen_US
dc.publisherInstitute of Electrical and Electronics Engineersen_US
dc.rights© 2024 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.en_US
dc.rightsThe following publication Z. Bian, X. Yan, J. Zhang, M. L. Yiu and B. Tang, "QSRP: Efficient Reverse k−Ranks Query Processing on High-Dimensional Embeddings," 2024 IEEE 40th International Conference on Data Engineering (ICDE), Utrecht, Netherlands, 2024, pp. 4614-4627 is available at https://doi.org/10.1109/ICDE60146.2024.00351.en_US
dc.subjectEmbedding modelen_US
dc.subjectHigh dimensional dataen_US
dc.subjectProduct-oriented recommendationen_US
dc.subjectRanking queryen_US
dc.subjectReverse k-ranksen_US
dc.titleQSRP : efficient reverse k−ranks query processing on high-dimensional embeddingsen_US
dc.typeConference Paperen_US
dc.identifier.spage4614en_US
dc.identifier.epage4627en_US
dc.identifier.doi10.1109/ICDE60146.2024.00351en_US
dcterms.abstractEmbedding models represent users and products as high-dimensional embedding vectors and are widely used for recommendation. In this paper, we study the reverse k-ranks query, which finds the users that are the most interested in a product and has many applications including product promotion, targeted advertising, and market analysis. As reverse k-ranks solutions for low dimensionality (e.g., trees) fail for the high-dimensional embeddings generated by embedding models, we propose the QSRP framework. QSRP precomputes the score table between all user and product embeddings to facilitate pruning and refinement at query time. As the score table is usually large, QSRP samples some of its columns as the index to fit in memory. To tackle the problem that naive uniform sampling results in poor pruning effect, we propose query-aware sampling, which conducts sampling by explicitly maximizing the pruning effect for a set of sample queries. Moreover, we introduce regression-based pruning, which fits cheap linear functions to predict the bounds used for pruning. We also design techniques to build the index with limited memory, reduce index building time, and handle updates. We evaluate QSRP under various configurations and compare with state-of-the-art baselines. The results show that QSRP achieves shorter query time than the baselines in all cases, and the speedup is usually over 100x.en_US
dcterms.accessRightsopen accessen_US
dcterms.bibliographicCitation2024 IEEE 40th International Conference on Data Engineering, 13-17 May 2024, Utrecht, Netherlands, p. 4614-4627en_US
dcterms.issued2024-
dc.relation.conferenceInternational Conference on Data Engineering [ICDE]en_US
dc.description.validate202408 bcchen_US
dc.description.oaAuthor’s Originalen_US
dc.identifier.FolderNumbera2990-
dc.identifier.SubFormID49084-
dc.description.fundingSourceRGCen_US
dc.description.pubStatusPublisheden_US
dc.description.oaCategoryGreen (AO)en_US
Appears in Collections:Conference Paper
Files in This Item:
File Description SizeFormat 
Bian_QSRP_Efficient_Reverse.pdfPreprint version1.51 MBAdobe PDFView/Open
Open Access Information
Status open access
File Version Author’s Original
Access
View full-text via PolyU eLinks SFX Query
Show simple item record

Page views

89
Citations as of Apr 14, 2025

Downloads

94
Citations as of Apr 14, 2025

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.