Mining and ranking generators of sequential patterns

Lo, D; Khoo, SC; Li, J

Mining and ranking generators of sequential patterns

Lo, D Khoo, SC Li, J

Permalink

Publication Type:: Conference Proceeding
Citation:: Society for Industrial and Applied Mathematics - 8th SIAM International Conference on Data Mining 2008, Proceedings in Applied Mathematics 130, 2008, 2 pp. 553 - 564
Issue Date:: 2008-10-01

Closed Access

	Filename	Description	Size
	2013003012OK.pdf	Published version	1.2 MB	Adobe PDF	View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Lo, D	en_US
dc.contributor.author	Khoo, SC	en_US
dc.contributor.author	Li, J https://orcid.org/0000-0003-1833-7413	en_US
dc.date.issued	2008-10-01	en_US
dc.identifier.citation	Society for Industrial and Applied Mathematics - 8th SIAM International Conference on Data Mining 2008, Proceedings in Applied Mathematics 130, 2008, 2 pp. 553 - 564	en_US
dc.identifier.isbn	9781605603179	en_US
dc.identifier.uri	http://hdl.handle.net/10453/32050
dc.description.abstract	Sequential pattern mining first proposed by Agrawal and Srikant has received intensive research due to its wide range applicability in many real-life domains. Various improvements have been proposed which include mining a closed set of sequential patterns. Sequential patterns supported by the same sequences in the database can be considered as belonging to an equivalence class. Each equivalence class contains patterns partially-ordered by sub-sequence relationship and having the same support. Within an equivalence class, the set of maximal and minimal patterns are referred to as closed patterns and generators respectively. Generators used together with closed patterns can provide additional Information which closed patterns alone are not able to provide. Also, as generators are the minimal members, they are preferable over closed patterns for model selection and classification based on the Minimum Description Length (MDL) principle. Several algorithms have been proposed for mining closed sequential patterns, but none so far for mining sequential generators. This paper fills this research gap by investigating properties of sequential generators and proposing an algorithm to efficiently mine sequential generators. The algorithm works on a three-step process of search space compaction, non-generator pruning and a final filtering step. We also introduce ranking of mined generators and propose mining of a unique generator per equivalence class. Performance study has been conducted on various synthetic and real benchmark datasets. They show that mining generators can be as fast as mining closed patterns even at low support thresholds. Copyright © by SIAM.	en_US
dc.relation.ispartof	Society for Industrial and Applied Mathematics - 8th SIAM International Conference on Data Mining 2008, Proceedings in Applied Mathematics 130	en_US
dc.title	Mining and ranking generators of sequential patterns	en_US
dc.type	Conference Proceeding
utslib.citation.volume	2	en_US
utslib.for	0804 Data Format	en_US
utslib.for	0801 Artificial Intelligence and Image Processing	en_US
dc.location.activity	Atlanta
pubs.embargo.period	Not known	en_US
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Strength - AAI - Advanced Analytics Institute Research Centre
pubs.organisational-group	/University of Technology Sydney/Strength - CHT - Health Technologies
utslib.copyright.status	closed_access
pubs.publication-status	Published	en_US
pubs.volume	2	en_US

Abstract:

Sequential pattern mining first proposed by Agrawal and Srikant has received intensive research due to its wide range applicability in many real-life domains. Various improvements have been proposed which include mining a closed set of sequential patterns. Sequential patterns supported by the same sequences in the database can be considered as belonging to an equivalence class. Each equivalence class contains patterns partially-ordered by sub-sequence relationship and having the same support. Within an equivalence class, the set of maximal and minimal patterns are referred to as closed patterns and generators respectively. Generators used together with closed patterns can provide additional Information which closed patterns alone are not able to provide. Also, as generators are the minimal members, they are preferable over closed patterns for model selection and classification based on the Minimum Description Length (MDL) principle. Several algorithms have been proposed for mining closed sequential patterns, but none so far for mining sequential generators. This paper fills this research gap by investigating properties of sequential generators and proposing an algorithm to efficiently mine sequential generators. The algorithm works on a three-step process of search space compaction, non-generator pruning and a final filtering step. We also introduce ranking of mined generators and propose mining of a unique generator per equivalence class. Performance study has been conducted on various synthetic and real benchmark datasets. They show that mining generators can be as fast as mining closed patterns even at low support thresholds. Copyright © by SIAM.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/32050