Mining closed sequences with constraint based on BIDE algorithm

Author

Shyamala, S. ; Sathya, T.

Author_Institution

Dept. of Comput. Sci., Sona Coll. of Technol., Salem, India

fYear

2012

fDate

10-12 Jan. 2012

Firstpage

1

Lastpage

5

Abstract

Mining sequential pattern is one of the common data mining task for many real-life applications. Previous existing algorithm such as CAMLS(Constraint-based Apriori Algorithm for Mining Long Sequences) mines the complete set of frequent sequences(Long) satisfying a min-sup threshold in a sequence. However, mining long sequences will generate an explosive number of frequent sequences, which is prohibitively costly in both run time and space storage. In this paper, we propose to improve CAMLS algorithm to produce only for closed sequences. Instead of mining full set of sequences, we plan to mine only short(closed) sequences. i.e., those containing, no super sequences with same support. Our motivation is to mine closed sequences from long sequences using BIDE algorithm with improved CAMLS algorithm and make the pruning strategy even more efficient. BIDE is an efficient algorithm for mining closed sequences which works under without candidate-maintenance and test paradigm.

Keywords

data mining; BIDE algorithm; CAMLS algorithm; bidirectional extension; closed sequence mining; constraint-based apriori algorithm for mining long sequences; data mining; pruning strategy; sequential pattern mining; Algorithm design and analysis; Computers; Data mining; Databases; Explosives; Informatics; Machine learning algorithms; Constraint-based mining; closed sequences; frequent patterns; long sequences; sequence database;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer Communication and Informatics (ICCCI), 2012 International Conference on

Conference_Location

Coimbatore

Print_ISBN

978-1-4577-1580-8

Type

conf

DOI

10.1109/ICCCI.2012.6158826

Filename

6158826