BotHunter
Abstract
The behaviors and characteristics of humans, bots, and cyborgs can discriminate them from one another. Cyborgs generate more tweets" than both humans and bots because they post manually as well as automatically. Bots tweet more frequently in their active periods b"ut due to long inactivity periods or suspension, they generate less tweets in their lifetime. Moreover, human connections are usuall""y reciprocal; i.e. the number of friends and followers for a given user should be similar. Bots, however, follow many users but rece""ive far fewer followers in response. In order to curtail the abusive activity of these types of bots, Twitter imposes a limit on the" friend to follower ratio. Sophisticated botsattempt to avoid this trap by unfollowing users who do not follow them back. Additiona"lly, different types of bots exhibit differences in the regularity of their tweeting behavior. While humans are more active on weekd""ays, bots have sustained activity levels throughout the week. Cyborgs are quite active onMondays as their controllers resume their" work cycle.We propose a system for real-time categorization of bots as they appear in data collection jobs on TweetTracker. Fig. 1 provides an initial architecture for a bot characterization platform using ASUs TweetTracker system. As tweets are captured by TweetTracker they are fed into a module which extracts six set of features. These features describe the content and structure of the t"weet, its author, and the network of users around it. These features are fed into a series of pre-trained classifiers that provide s""tatistical measures to guide bot characterization for research on intelligent analysis, evolution tracking, and impact analysis. We" will investigate how bots can be investigated with knowncharacteristics by designing passive bots to assist our research.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 23, 2018
- Source ID
- N000141812108
Entities
People
- Kathleen Carley
Organizations
- Massachusetts Institute of Technology
- Office of Naval Research
- United States Navy