电子学报
電子學報
전자학보
ACTA ELECTRONICA SINICA
2015年
8期
1545-1554
,共10页
数据挖掘%完全加权关联规则%负关联规则%频繁项集
數據挖掘%完全加權關聯規則%負關聯規則%頻繁項集
수거알굴%완전가권관련규칙%부관련규칙%빈번항집
data mining%all-weighted association rule%negative association rule%frequent itemset
本文提出一种基于项权值变化的完全加权正负关联规则挖掘算法,解决了基于项权值变化的负模式挖掘问题.该算法考虑项权值依赖于事务记录的特点,采用新的项集剪枝方法和模式评价框架,通过项集的项内权值比和维数比的简单计算和比较,挖掘有效的完全加权正负关联规则.实验结果表明,与现有无加权正负关联规则挖掘算法比较,本文算法能避免无效的模式出现,其挖掘时间和候选项集数量明显减少,减幅最大分别可达94.09%和88.16%.
本文提齣一種基于項權值變化的完全加權正負關聯規則挖掘算法,解決瞭基于項權值變化的負模式挖掘問題.該算法攷慮項權值依賴于事務記錄的特點,採用新的項集剪枝方法和模式評價框架,通過項集的項內權值比和維數比的簡單計算和比較,挖掘有效的完全加權正負關聯規則.實驗結果錶明,與現有無加權正負關聯規則挖掘算法比較,本文算法能避免無效的模式齣現,其挖掘時間和候選項集數量明顯減少,減幅最大分彆可達94.09%和88.16%.
본문제출일충기우항권치변화적완전가권정부관련규칙알굴산법,해결료기우항권치변화적부모식알굴문제.해산법고필항권치의뢰우사무기록적특점,채용신적항집전지방법화모식평개광가,통과항집적항내권치비화유수비적간단계산화비교,알굴유효적완전가권정부관련규칙.실험결과표명,여현유무가권정부관련규칙알굴산법비교,본문산법능피면무효적모식출현,기알굴시간화후선항집수량명현감소,감폭최대분별가체94.09%화88.16%.
This paper proposes a mining algorithm of all-weighted positive and negative association rules based on dynamic item weight,which can solve the problems of negative patterns mining based on dynamic item weight.This algorithm took the dy-namic item weight dependent on transaction records into consideration,and adopted the itemset pruning method and pattern evalua-tion framework so as to discover effective all-weighted positive &negative association rules via simple calculation and comparison of weight ratio and dimension ratio from the itemset.The experimental results show that this algorithm can prevent ineffective patterns, which makes the maximal declines of the mining time and number of the candidate itemsets by up to 94.09% and 88.16% respec-tively compared with the existing unweighted positive and negative association rule mining algorithms.