基于自适应高斯混合模型的数据库基数估计方法

打开文本图片集

关键词：查询优化；基数估计；自适应高斯混合模型；自回归模型；偏差校正中图分类号：TP311.13 文献标志码：A 文章编号：1001-3695（2026）04-024-1171-09doi：10.19734/j.issn.1001-3695.2025.07.0292

Database cardinalityestimation method based on adaptive Gaussian mixture model

Li Hao，Liu Mengchi ，Zou Ruiji，Liu Mingkai （SchoolofComputer Science，SouthChina Normal University，Guangzhou 51oooo，China）

Abstract：Cardinalityestimationisa criticalcomponent ofdatabase queryoptimization，where itsaccuracy directlyimpacts theexecution eficiencyofqueryplans.Deepautoregresive model-basedcardinalityestimators havedemonstratedremarkable accuracyinpriorstudies.However，theystrugletocapturedatadistributionpaternswhenhandlinglarge-domaincontinuous atributes，whichleadtosignificantperformancedegradation.Toadessthesechalenges，thispaperproposedanovelcardinalityestimatorbasedonanadaptive Gaussianmixturemodel，calledAGCardIt firstdynamicallyadjusted thenumberand parameters ofGaussancomponents toadaptivelyfitthedatadistributionofcontiuous atributes，therebyreducingthedomain scale.Subsequently，AGCard employedabiascorrctionalgorithmtocompensate fortheestimationdeviations introducedby the progresivesampling process whileavoiding additional computationaloverhead.Extensive experiments on threereal-world datasets （including WISDM）demonstrate thatthe proposed method outperforms existing mainstream baselines in terms of estimationaccuracy，inferencelatency，andstorageoverhead.Theresultsconfirmtheefectiveness oftheadaptiveGaussianmixture model and the bias correction algorithm.

Key words：queryoptimization；cardinalityestimation;adaptive Gaussanmixture model;；autoregressive model;bias correc tion

0引言

基数估计作为数据库管理系统查询优化的关键环节，在系统执行查询前估计满足查询语句的元组数量，对提升数据库系统的性能至关重要。（剩余22128字）