Temporal Patterns in Bot Activities
Authors:
Open Access
- 1 January 2017
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1601-1606
Abstract
Correlated or synchronized bots commonly exist in social media sites such as Twitter. Bots work towards gaining human followers, participating in campaigns, and engaging in unethical activities such as spamming and false click generation. In this paper, we perform temporal pattern mining on bot activities in Twitter. We discover motifs (repeating behavior), discords (anomalous behavior), joins, bursts and dynamic clusters in activities of Twitter bots, and explain the significance of these temporal patterns in gaining competitive advantage over humans. Our analysis identifies a small set of indicators that separates bots from humans with high precision.Keywords
This publication has 11 references indexed in Scilit:
- DeBot: Twitter Bot Detection via Warped Correlation2016
- AWarp: Fast Warping Distance for Sparse Time Series2016
- On URL Changes and Handovers in Social Media2016
- Identifying Correlated Bots in Twitter2016
- The rise of social botsCommunications of the ACM, 2016
- The DARPA Twitter Bot ChallengeComputer, 2016
- Dual-Domain Hierarchical Classification of Phonetic Time Series2014
- Time Series Join on Subsequence Correlation2014
- Groundhog day2013
- Exact Discovery of Time Series Motifs2009