Sparsification of influence networks

Michael Mathioudakis, Francesco Bonchi, Carlos Castillo, Aristides Gionis, Antti Ukkonen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

107 Citations (Scopus)

Abstract

We present Spine, an efficient algorithm for finding the "backbone" of an influence network. Given a social graph and a log of past propagations, we build an instance of the independent-cascade model that describes the propagations. We aim at reducing the complexity of that model, while preserving most of its accuracy in describing the data. We show that the problem is inapproximable and we present an optimal, dynamic-programming algorithm, whose search space, albeit exponential, is typically much smaller than that of the brute force, exhaustive-search approach. Seeking a practical, scalable approach to sparsification, we devise Spine, a greedy, efficient algorithm with practically little compromise in quality. We claim that sparsification is a fundamental datareduction operation with many applications, ranging from visualization to exploratory and descriptive data analysis. As a proof of concept, we use Spine on real-world datasets, revealing the backbone of their influence-propagation networks. Moreover, we apply Spine as a pre-processing step for the influence-maximization problem, showing that computations on sparsified models give up little accuracy, but yield significant improvements in terms of scalability.

Original languageEnglish
Title of host publicationProceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
Pages529-537
Number of pages9
DOIs
Publication statusPublished - 16 Sep 2011
Externally publishedYes
Event17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD'11 - San Diego, CA, United States
Duration: 21 Aug 201124 Aug 2011

Other

Other17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD'11
CountryUnited States
CitySan Diego, CA
Period21/8/1124/8/11

Fingerprint

Dynamic programming
Scalability
Visualization
Processing

Keywords

  • Influence
  • Propagation
  • Social networks

ASJC Scopus subject areas

  • Software
  • Information Systems

Cite this

Mathioudakis, M., Bonchi, F., Castillo, C., Gionis, A., & Ukkonen, A. (2011). Sparsification of influence networks. In Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 529-537) https://doi.org/10.1145/2020408.2020492

Sparsification of influence networks. / Mathioudakis, Michael; Bonchi, Francesco; Castillo, Carlos; Gionis, Aristides; Ukkonen, Antti.

Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2011. p. 529-537.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Mathioudakis, M, Bonchi, F, Castillo, C, Gionis, A & Ukkonen, A 2011, Sparsification of influence networks. in Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. pp. 529-537, 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD'11, San Diego, CA, United States, 21/8/11. https://doi.org/10.1145/2020408.2020492
Mathioudakis M, Bonchi F, Castillo C, Gionis A, Ukkonen A. Sparsification of influence networks. In Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2011. p. 529-537 https://doi.org/10.1145/2020408.2020492
Mathioudakis, Michael ; Bonchi, Francesco ; Castillo, Carlos ; Gionis, Aristides ; Ukkonen, Antti. / Sparsification of influence networks. Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2011. pp. 529-537
@inproceedings{97b14a1518a14bf682f4c3fa905ac1a3,
title = "Sparsification of influence networks",
abstract = "We present Spine, an efficient algorithm for finding the {"}backbone{"} of an influence network. Given a social graph and a log of past propagations, we build an instance of the independent-cascade model that describes the propagations. We aim at reducing the complexity of that model, while preserving most of its accuracy in describing the data. We show that the problem is inapproximable and we present an optimal, dynamic-programming algorithm, whose search space, albeit exponential, is typically much smaller than that of the brute force, exhaustive-search approach. Seeking a practical, scalable approach to sparsification, we devise Spine, a greedy, efficient algorithm with practically little compromise in quality. We claim that sparsification is a fundamental datareduction operation with many applications, ranging from visualization to exploratory and descriptive data analysis. As a proof of concept, we use Spine on real-world datasets, revealing the backbone of their influence-propagation networks. Moreover, we apply Spine as a pre-processing step for the influence-maximization problem, showing that computations on sparsified models give up little accuracy, but yield significant improvements in terms of scalability.",
keywords = "Influence, Propagation, Social networks",
author = "Michael Mathioudakis and Francesco Bonchi and Carlos Castillo and Aristides Gionis and Antti Ukkonen",
year = "2011",
month = "9",
day = "16",
doi = "10.1145/2020408.2020492",
language = "English",
isbn = "9781450308137",
pages = "529--537",
booktitle = "Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining",

}

TY - GEN

T1 - Sparsification of influence networks

AU - Mathioudakis, Michael

AU - Bonchi, Francesco

AU - Castillo, Carlos

AU - Gionis, Aristides

AU - Ukkonen, Antti

PY - 2011/9/16

Y1 - 2011/9/16

N2 - We present Spine, an efficient algorithm for finding the "backbone" of an influence network. Given a social graph and a log of past propagations, we build an instance of the independent-cascade model that describes the propagations. We aim at reducing the complexity of that model, while preserving most of its accuracy in describing the data. We show that the problem is inapproximable and we present an optimal, dynamic-programming algorithm, whose search space, albeit exponential, is typically much smaller than that of the brute force, exhaustive-search approach. Seeking a practical, scalable approach to sparsification, we devise Spine, a greedy, efficient algorithm with practically little compromise in quality. We claim that sparsification is a fundamental datareduction operation with many applications, ranging from visualization to exploratory and descriptive data analysis. As a proof of concept, we use Spine on real-world datasets, revealing the backbone of their influence-propagation networks. Moreover, we apply Spine as a pre-processing step for the influence-maximization problem, showing that computations on sparsified models give up little accuracy, but yield significant improvements in terms of scalability.

AB - We present Spine, an efficient algorithm for finding the "backbone" of an influence network. Given a social graph and a log of past propagations, we build an instance of the independent-cascade model that describes the propagations. We aim at reducing the complexity of that model, while preserving most of its accuracy in describing the data. We show that the problem is inapproximable and we present an optimal, dynamic-programming algorithm, whose search space, albeit exponential, is typically much smaller than that of the brute force, exhaustive-search approach. Seeking a practical, scalable approach to sparsification, we devise Spine, a greedy, efficient algorithm with practically little compromise in quality. We claim that sparsification is a fundamental datareduction operation with many applications, ranging from visualization to exploratory and descriptive data analysis. As a proof of concept, we use Spine on real-world datasets, revealing the backbone of their influence-propagation networks. Moreover, we apply Spine as a pre-processing step for the influence-maximization problem, showing that computations on sparsified models give up little accuracy, but yield significant improvements in terms of scalability.

KW - Influence

KW - Propagation

KW - Social networks

UR - http://www.scopus.com/inward/record.url?scp=80052654411&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80052654411&partnerID=8YFLogxK

U2 - 10.1145/2020408.2020492

DO - 10.1145/2020408.2020492

M3 - Conference contribution

SN - 9781450308137

SP - 529

EP - 537

BT - Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

ER -