A Bayesian Network Model for Interesting Itemsets

Jaroslav Fowkes; Charles Sutton

doi:10.1007/978-3-319-46227-1_26

A Bayesian Network Model for Interesting Itemsets

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Mining itemsets that are the most interesting under a statistical model of the underlying data is a commonly used and well-studied technique for exploratory data analysis, with the most recent interestingness models exhibiting state of the art performance. Continuing this highly promising line of work, we propose the first, to the best of our knowledge, generative model over itemsets, in the form of a Bayesian network, and an associated novel measure of interestingness. Our model is able to efficiently infer interesting itemsets directly from the transaction database using structural EM, in which the E-step employs the greedy approximation to weighted set cover. Our approach is theoretically simple, straightforward to implement, trivially parallelizable and retrieves
itemsets whose quality is comparable to, if not better than, existing state
of the art algorithms as we demonstrate on several real-world datasets.

Original language	English
Title of host publication	The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery (ECML-PKDD 2016)
Place of Publication	Riva del Garda, Italy
Publisher	Springer
Pages	410-425
Number of pages	16
ISBN (Electronic)	978-3-319-46227-1
ISBN (Print)	978-3-319-46226-4
DOIs	https://doi.org/10.1007/978-3-319-46227-1_26
Publication status	Published - 4 Sept 2016
Event	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2016 - Riva del Garda, Italy Duration: 19 Sept 2016 → 23 Sept 2016 http://www.ecmlpkdd2016.org/

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer, Cham
Volume	9852
ISSN (Print)	0302-9743

Conference

Conference	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2016
Abbreviated title	ECML-PKDD 2016
Country/Territory	Italy
City	Riva del Garda
Period	19/09/16 → 23/09/16
Internet address	http://www.ecmlpkdd2016.org/

Access to Document

10.1007/978-3-319-46227-1_26

pkdd2016-2Accepted author manuscript, 467 KB

https://link.springer.com/chapter/10.1007/978-3-319-46227-1_26

Statistical Natural Language Processing Methods for Computer Program Source Code
Sutton, C. (Principal Investigator)
EPSRC
1/10/13 → 31/03/17
Project: Research

Cite this

@inproceedings{d003605313cc455eb167f0921bd2830b,

title = "A Bayesian Network Model for Interesting Itemsets",

abstract = "Mining itemsets that are the most interesting under a statistical model of the underlying data is a commonly used and well-studied technique for exploratory data analysis, with the most recent interestingness models exhibiting state of the art performance. Continuing this highly promising line of work, we propose the first, to the best of our knowledge, generative model over itemsets, in the form of a Bayesian network, and an associated novel measure of interestingness. Our model is able to efficiently infer interesting itemsets directly from the transaction database using structural EM, in which the E-step employs the greedy approximation to weighted set cover. Our approach is theoretically simple, straightforward to implement, trivially parallelizable and retrievesitemsets whose quality is comparable to, if not better than, existing stateof the art algorithms as we demonstrate on several real-world datasets.",

author = "Jaroslav Fowkes and Charles Sutton",

year = "2016",

month = sep,

day = "4",

doi = "10.1007/978-3-319-46227-1_26",

language = "English",

isbn = "978-3-319-46226-4",

series = "Lecture Notes in Computer Science ",

publisher = "Springer",

pages = "410--425",

booktitle = "The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery (ECML-PKDD 2016)",

address = "United Kingdom",

note = "European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2016, ECML-PKDD 2016 ; Conference date: 19-09-2016 Through 23-09-2016",

url = "http://www.ecmlpkdd2016.org/",

}

Fowkes, J & Sutton, C 2016, A Bayesian Network Model for Interesting Itemsets. in The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery (ECML-PKDD 2016). Lecture Notes in Computer Science , vol. 9852, Springer, Riva del Garda, Italy, pp. 410-425, European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2016, Riva del Garda, Italy, 19/09/16. https://doi.org/10.1007/978-3-319-46227-1_26

A Bayesian Network Model for Interesting Itemsets. / Fowkes, Jaroslav; Sutton, Charles.
The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery (ECML-PKDD 2016). Riva del Garda, Italy: Springer, 2016. p. 410-425 (Lecture Notes in Computer Science ; Vol. 9852).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - A Bayesian Network Model for Interesting Itemsets

AU - Fowkes, Jaroslav

AU - Sutton, Charles

PY - 2016/9/4

Y1 - 2016/9/4

N2 - Mining itemsets that are the most interesting under a statistical model of the underlying data is a commonly used and well-studied technique for exploratory data analysis, with the most recent interestingness models exhibiting state of the art performance. Continuing this highly promising line of work, we propose the first, to the best of our knowledge, generative model over itemsets, in the form of a Bayesian network, and an associated novel measure of interestingness. Our model is able to efficiently infer interesting itemsets directly from the transaction database using structural EM, in which the E-step employs the greedy approximation to weighted set cover. Our approach is theoretically simple, straightforward to implement, trivially parallelizable and retrievesitemsets whose quality is comparable to, if not better than, existing stateof the art algorithms as we demonstrate on several real-world datasets.

AB - Mining itemsets that are the most interesting under a statistical model of the underlying data is a commonly used and well-studied technique for exploratory data analysis, with the most recent interestingness models exhibiting state of the art performance. Continuing this highly promising line of work, we propose the first, to the best of our knowledge, generative model over itemsets, in the form of a Bayesian network, and an associated novel measure of interestingness. Our model is able to efficiently infer interesting itemsets directly from the transaction database using structural EM, in which the E-step employs the greedy approximation to weighted set cover. Our approach is theoretically simple, straightforward to implement, trivially parallelizable and retrievesitemsets whose quality is comparable to, if not better than, existing stateof the art algorithms as we demonstrate on several real-world datasets.

U2 - 10.1007/978-3-319-46227-1_26

DO - 10.1007/978-3-319-46227-1_26

M3 - Conference contribution

SN - 978-3-319-46226-4

T3 - Lecture Notes in Computer Science

SP - 410

EP - 425

BT - The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery (ECML-PKDD 2016)

PB - Springer

CY - Riva del Garda, Italy

T2 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2016

Y2 - 19 September 2016 through 23 September 2016

ER -

A Bayesian Network Model for Interesting Itemsets

Abstract

Publication series

Conference

Access to Document

Fingerprint

Projects

Statistical Natural Language Processing Methods for Computer Program Source Code

Cite this