Tool Developed for Identifying Bots on Twitter
With the prevalence of social media systems and people's growing reliance on them for information, it is not surprising that programs have been created to automatically share information. While these bots can be useful in some situations, they could also be deployed with malicious intent and spread misinformation. Researchers funded by the NSF and Department of Defense at Indiana University have created the program, BotOrNot to analyze and report on if a Twitter account is human or not.
You may be wondering why the Department of Defense is interested in whether Twitter accounts being run by people or bots. It is directly because of the potential deception social networks could rapidly spread during times of emergency, the effects misinformation can have on stock markets, and more. To identify when an account is a bot or not, the system considers over 1000 features of an account, including its friendship network, content, and temporal information. It then applies these data to predict what the account is, and according to one evaluation method BotOrNot scores 0.95 out of a perfect score of 1.0.
Another interesting aspect of this research is applying it to determine just how many bots there are. We do know a great many exist and that most are benign, but exactly how many are out there and how many may have malicious purposes is unknown.
Source: Indiana University