从图数据库中挖掘频繁跳跃模式.pdf

ISSN 1000-9825．CODENRUXUEW E—mail：jos@isc嬲．ac．cn 乃m口，可S够MM，V01．2l，No．10，Octob盯20lO，即．2477-2493 http：／／wwwJos．o唱．cn doi：lO．3724／SPJ．1001．2010．0383l O rights瑚crvcd． by如JffmfP∥s够M坤，f拓∞加嚣P一翻如删旷f积c嚣．All 从图数据库中挖掘频繁跳跃模式幸 刘 勇+，李建中，高宏 (哈尔滨工业大学计算机科学与技术学院，黑龙江哈尔滨 150001) Patternsftom Databases MiningFrequentJunlp Graph LIU LI GA0 Yong+， Ji船-Zhong， Hong I衄伽lte (SchoolofComput盯Sci∞cc柚dT∞hnology，HarbinofT∞hnology，Harbin15000l，China) +co肿sponding卸thof：E—mil：liuy∞9123456@hit．edu．∞ Liu ftom Y，LiJZ，Gao H．Miningftequentjumppatternsgraph 2477-2493．hnp：，翩．jos．org．c“1000一9825，3831．h恤 Abstract： on havebeen n啪ber Manyalgorithmssubgr叩hmining proposed．Howeverthe betoo tobe whcnthe these suppon generatedby aIgorithmsmay large e仃ectiVelyexplorcdbyusers，especially thresholdislow．Inthis new of datab·asesis p印er，a problemmining仃equentjumppanems丘Dm口叩h proposed． reducethenumberof still c蚰dramatically pattems粕dc印nH’e Miningfrequentj硼叩panems output鲫h arerobust noiseand indata． imeresting伊aphpattems．Futhe彻ore，jumppattems agai璐t dynamicchanges is duetothe associated as However，thisproblemchallengingunderlyingcomplexitywith矗equentsub鲫hmining wcllastheabsenceof for the noVel Aprioripropenyj啪ppanems．Byexploringpr叩ertiesofjumppattems，铆o e疗Icctive