BA/MA: Building (Social) Bots
Social bots are automated or semi-automated entities active in online social networks or communities. Their distinguishing feature is that they pretend to be legitimate human users. In order to do so, social bots are programmed to (1) mimic human users in temporal activity patterns, (2) create a network of friends and followers, and (3) produce credible and intelligent content.
The focus of this thesis is on the conceptualization and creation of bots that mimic human users and successfully interact with other users on the network Twitter through the Twitter API. The aim will be to mimic the capabilities of social bots with the crucial difference that your bot(s) will not pretend to be human users. Depending on your interests and abilities, the thesis can be of variable focus and complexity. Among the possible topics are:
The implementation of basic chat capabilities (for example, chat capabilities can be implemented using AIML or the Python library ChatterBot);
The automatic distribution of specific content from the web (for example, news from specific outlets);
The comparison, test, and evaluation of several bots with different parameters (for example, to test different methods to mimic human users’ temporal activity patterns);
Bot-to-bot interaction 1: You develop more than one bot and consider how they may interact with one another and other users (for example, by having a “conversation”).
Bot-to-bot interaction 2: Your bot(s) can focus on interactions with other user accounts that have been identified as potential bots through https://botometer.iuni.iu.edu/ (for example, to conduct automatized tests and thereby verify whether these accounts are bots or not).
By conceptualizing and developing bots that are active on social networks, the thesis contributes to a broader research agenda that investigates the influence of social bots on online political discourse. A central aim of this broader research is to be able to identify social bots and mitigate the negative influence of social bots on online political discourse, but in order do do so, a good understanding about their abilities and various designs is necessary. To further such an understanding, it is important to learn what human user activities are relatively simple to mimic for bots and what human user activities are difficult if not impossible to mimic, for which your work will be important.
Theses about this topic would be supervized by Ole Pütz. Please contact Ole for further information.