An Improved Top-Down Data Mining Algorithm for Long Frequents

Fang, Gang; Liu, Yulu; Xiong, Jiang; Hong, Ying

doi:10.1109/aici.2009.315

Cited by 2 publications

(4 citation statements)

References 5 publications

(4 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Namely, it uses digit logical operation to generate k-candidate frequent itemsets of digit form by (k+1)-digit non-frequent itemsets. The algorithm adopts search strategy which is similar to B_UDMA as in [7] and ITDASN as in [8]. The course of generating is expressed as follows:…”

Section: B the Methods Of Generating Candidate Frequent Ncsmentioning

confidence: 99%

The application of a top-down algorithm in neighboring class set mining

Fang

Cheng-sheng

Xiong

et al. 2010

2010 IEEE International Conference on Intelligent Systems and Knowledge Engineering

Self Cite

View full text Add to dashboard Cite

This paper focuses on character of present frequent neighboring class set mining algorithms which is suitable for mining short frequent neighboring class set, and introduces a top-down algorithm in frequent neighboring class set mining. This algorithm is suitable for mining long frequent neighboring class set in large spatial data according to top-down strategy, and it creates digital database of neighboring class set via neighboring class bit sequence. The algorithm generates candidate frequent neighboring class set via top-down search strategy, namely, it gains k-neighboring class set as candidate frequent items by computing k-subset of (k+1)-non frequent neighboring class set. The mining algorithm computes support of candidate frequent neighboring class set by digit logical operation. The algorithm improves mining efficiency through these two methods. The result of experiment indicates that the algorithm is faster and more efficient than present algorithms when mining long frequent neighboring class set in large spatial data.

show abstract

Section: B the Methods Of Generating Candidate Frequent Ncsmentioning

confidence: 99%

The application of a top-down algorithm in neighboring class set mining

Fang

Cheng-sheng

Xiong

et al. 2010

2010 IEEE International Conference on Intelligent Systems and Knowledge Engineering

Self Cite

View full text Add to dashboard Cite

show abstract

“…According to the content of reference [4] and [5], we knew that ARNBSN is more efficient than B-Apriori when mining short frequent itemsets, ITDASN is only suitable for mining long frequent itemsets. Hence, here we only compare ALOMSN with ARNBSN and ITDASN.…”

Section: Comparing the Capability Of Algorithmsmentioning

confidence: 99%

“…B-Apriori as in [2] is faster than Apriori for binary logic operation, and B-ARDSM [3] is also faster than IDMFIA when double mining, and although ARNBSN as in [4] is faster than B-Apriori, it is only suitable for mining short frequent itemsets. ITDASN as in [5] is suitable for mining long frequent itemsets, but there are still redundant candidate frequent itemsets and repeated computing when mining algorithm generates L-candidate frequent itemsets of (L+1)-non frequent itemsets. And so, this paper proposes an algorithm of locating order mining based on sequence number, denoted by ALOMSN, which is suitable for mining long frequent itemsets, and is faster and more efficient than ITDASN as in [5].…”

Section: Introductionmentioning

confidence: 99%

“…ITDASN as in [5] is suitable for mining long frequent itemsets, but there are still redundant candidate frequent itemsets and repeated computing when mining algorithm generates L-candidate frequent itemsets of (L+1)-non frequent itemsets. And so, this paper proposes an algorithm of locating order mining based on sequence number, denoted by ALOMSN, which is suitable for mining long frequent itemsets, and is faster and more efficient than ITDASN as in [5].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An algorithm of locating order mining based on sequence number

Fang

Hong

Xiong

et al. 2010

2010 International Conference on Machine Learning and Cybernetics

Self Cite

View full text Add to dashboard Cite

At present, existing association rules mining algorithms have redundant candidate frequent itemsets and repeated computing. This paper proposes an algorithm of locating order mining based on sequence number, which is suitable for mining long frequent itemsets. In order to fast search long frequent itemsets, the algorithm adopts not only traditional down search, but also the method of locating order of subset to generate candidate frequent itemsets. It has two aspects, which are different from traditional down search mining algorithm. One is that the algorithm need locate order of subsets of non frequent itemsets via down search. The other is that the algorithm uses character of attribute sequence number to compute support for only scanning database once. The algorithm may efficiently delete repeated L-candidate frequent itemsets generated by (L+1)-non frequent itemsets via locating subsets' order, whose efficiency is improved. The result of experiment indicates that the algorithm is suitable for mining long frequent itemsets, and it is faster and more efficient than present algorithms of mining long frequent itemsets. Definition 1 Binary Transaction (BT), there is a transaction T is expressed as character stringExample let I= {1, 2, 3, 4, 5, 6} be an itemsets, if a transaction is expressed as T i = {2, 5, 6}, and then BT i = (010011).Definition 2 Digital Transaction (DT), it is an integer, and its value is decimal integer of binary transaction.Example if BT= (010010), and then DT=18. Definition 3 Digital Item (DI), it is an integer, and it is the simplest digital transaction which only expresses an item attribute.Example let I= {i 1 , i 2 …i m }, and then DI 1 =1… DI m = 2 m-1 .Definition 4 Digital Transaction Length (DTL), it is an integer, it is equal to the sum of "1" in Binary Transaction.Definition 5 Suppose digital transaction of T 1 is denoted by DT 1 , digital transaction of T 2 is denoted by DT 2 . If T 1 ⊆ T 2 , and then DT 1 ⊆ DT 2 , DT 1 is regarded as subset of DT 2 , which is regarded as superset of DT 1 .Definition 6 Sequence Number (SN), it is a group of ordered number, here, these numbers may be repeated, and each number is called a sub-Item of sequence number.Example let SN= {46, 124, 65, 125, 79, 62, and 112} be a sequence number, thereinto, 124 is called a sub-Item of Sequence Number.Definition 7 Sub Item Dimension (SID), it is an integer, it is equal to the sum of "1" in binary code of sub-Item.Example, let 58 be a sub-Item, and then SID (58) = SID (111010) 2 =4.Definition 8 Sequence Number Dimension (SND), it is an integer, it is equal to the sum of items' SID contained by 403 2010 IEEE 978-1-4244-6527-9/10/$26.00 ©

show abstract

An Improved Top-Down Data Mining Algorithm for Long Frequents

Cited by 2 publications

References 5 publications

The application of a top-down algorithm in neighboring class set mining

The application of a top-down algorithm in neighboring class set mining

An algorithm of locating order mining based on sequence number

Contact Info

Product

Resources

About