Abstract

Online information propagates differently on the web, some of which can be viral. In this paper, first we introduce a simple standard deviation sigma levels based Tweet volume breakout definition, then we proceed to determine patterns of re-tweet network measures to predict whether a hashtag volume will breakout or not. We also developed a visualization tool to help trace the evolution of hashtag volumes, their underlying networks and both local and global network measures. We trained a random forest tree classifier to identify effective network measures for predicting hashtag volume breakouts. Our experiments showed that “local” network features, based on a fixed-sized sliding window, have an overall predictive accuracy of 76%, where as, when we incorporate “global” features that utilize all interactions up to the current period, then the overall predictive accuracy of a sliding window based breakout predictor jumps to 83%.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages3-12
Number of pages10
Volume9021
ISBN (Print)9783319162676
DOIs
StatePublished - 2015
Event8th International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction, SBP 2015 - Washington, United States
Duration: Mar 31 2015Apr 3 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9021
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other8th International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction, SBP 2015
CountryUnited States
CityWashington
Period3/31/154/3/15

Fingerprint

Classifiers
Visualization
Sliding Window
Experiments
Model
Random Forest
Standard deviation
Predictors
Jump
Classifier
Trace
Predict
Interaction
Experiment

Keywords

  • Diffusion networks
  • Hashtag volumes
  • Information diffusion
  • Prediction
  • Social networks

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Alzahrani, S., Alashri, S., Koppela, A. R., Davulcu, H., & Toroslu, I. (2015). A network-based model for predicting hashtag breakouts in twitter. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9021, pp. 3-12). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9021). Springer Verlag. https://doi.org/10.1007/978-3-319-16268-3_1

A network-based model for predicting hashtag breakouts in twitter. / Alzahrani, Sultan; Alashri, Saud; Koppela, Anvesh Reddy; Davulcu, Hasan; Toroslu, Ismail.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9021 Springer Verlag, 2015. p. 3-12 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9021).

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

Alzahrani, S, Alashri, S, Koppela, AR, Davulcu, H & Toroslu, I 2015, A network-based model for predicting hashtag breakouts in twitter. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9021, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9021, Springer Verlag, pp. 3-12, 8th International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction, SBP 2015, Washington, United States, 3/31/15. https://doi.org/10.1007/978-3-319-16268-3_1
Alzahrani S, Alashri S, Koppela AR, Davulcu H, Toroslu I. A network-based model for predicting hashtag breakouts in twitter. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9021. Springer Verlag. 2015. p. 3-12. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-16268-3_1
Alzahrani, Sultan ; Alashri, Saud ; Koppela, Anvesh Reddy ; Davulcu, Hasan ; Toroslu, Ismail. / A network-based model for predicting hashtag breakouts in twitter. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9021 Springer Verlag, 2015. pp. 3-12 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{f19d907ce59f4db5883a33172ef7ea98,
title = "A network-based model for predicting hashtag breakouts in twitter",
abstract = "Online information propagates differently on the web, some of which can be viral. In this paper, first we introduce a simple standard deviation sigma levels based Tweet volume breakout definition, then we proceed to determine patterns of re-tweet network measures to predict whether a hashtag volume will breakout or not. We also developed a visualization tool to help trace the evolution of hashtag volumes, their underlying networks and both local and global network measures. We trained a random forest tree classifier to identify effective network measures for predicting hashtag volume breakouts. Our experiments showed that “local” network features, based on a fixed-sized sliding window, have an overall predictive accuracy of 76{\%}, where as, when we incorporate “global” features that utilize all interactions up to the current period, then the overall predictive accuracy of a sliding window based breakout predictor jumps to 83{\%}.",
keywords = "Diffusion networks, Hashtag volumes, Information diffusion, Prediction, Social networks",
author = "Sultan Alzahrani and Saud Alashri and Koppela, {Anvesh Reddy} and Hasan Davulcu and Ismail Toroslu",
year = "2015",
doi = "10.1007/978-3-319-16268-3_1",
language = "English (US)",
isbn = "9783319162676",
volume = "9021",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "3--12",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - A network-based model for predicting hashtag breakouts in twitter

AU - Alzahrani, Sultan

AU - Alashri, Saud

AU - Koppela, Anvesh Reddy

AU - Davulcu, Hasan

AU - Toroslu, Ismail

PY - 2015

Y1 - 2015

N2 - Online information propagates differently on the web, some of which can be viral. In this paper, first we introduce a simple standard deviation sigma levels based Tweet volume breakout definition, then we proceed to determine patterns of re-tweet network measures to predict whether a hashtag volume will breakout or not. We also developed a visualization tool to help trace the evolution of hashtag volumes, their underlying networks and both local and global network measures. We trained a random forest tree classifier to identify effective network measures for predicting hashtag volume breakouts. Our experiments showed that “local” network features, based on a fixed-sized sliding window, have an overall predictive accuracy of 76%, where as, when we incorporate “global” features that utilize all interactions up to the current period, then the overall predictive accuracy of a sliding window based breakout predictor jumps to 83%.

AB - Online information propagates differently on the web, some of which can be viral. In this paper, first we introduce a simple standard deviation sigma levels based Tweet volume breakout definition, then we proceed to determine patterns of re-tweet network measures to predict whether a hashtag volume will breakout or not. We also developed a visualization tool to help trace the evolution of hashtag volumes, their underlying networks and both local and global network measures. We trained a random forest tree classifier to identify effective network measures for predicting hashtag volume breakouts. Our experiments showed that “local” network features, based on a fixed-sized sliding window, have an overall predictive accuracy of 76%, where as, when we incorporate “global” features that utilize all interactions up to the current period, then the overall predictive accuracy of a sliding window based breakout predictor jumps to 83%.

KW - Diffusion networks

KW - Hashtag volumes

KW - Information diffusion

KW - Prediction

KW - Social networks

UR - http://www.scopus.com/inward/record.url?scp=84925303347&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84925303347&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-16268-3_1

DO - 10.1007/978-3-319-16268-3_1

M3 - Conference contribution

SN - 9783319162676

VL - 9021

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 3

EP - 12

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

PB - Springer Verlag

ER -