A Q&A with a UCIPT Postdoctoral Researcher
Twitter has recently been cracking down on the Islamic State’s use of their platform. Have there been any studies related to student tweets that would enable you to identify students at risk for radicalization by ISIS or other terror groups?
Applying social media technology to identify online radicalization is an area that has attracted several researchers’ attention over past decade. There are numerous algorithms, techniques, and digital tools that have been proposed to counter and combat cyber-extremism and predict protest-related events in advance. These techniques include keyword searchers and machine learning techniques. However, the current accuracy to identify online radicalization still needs to be improved. You can read about the current methods to identify radicalization on social media here.
Does “hate speech” on Twitter stress students out? Or does it excite them?
Hate speech may make some people “excited,” but the person hearing/reading those remarks may also feel uncomfortable. As hate speech and cyberbullying become more common, particularly among young people, legislation and awareness campaigns have arisen to combat these problems. UCIPT has recently started to examine ways to identify online aggression and cyberbullying on microblogs. We will keep you posted once we have some results.
Are any public agencies, such as gun ownership databases, voter registration databases, motor vehicle authorities, or other agencies that interface with college students voluntarily asking for Twitter handles? What public or government resources exist for scholars or Freedom of Information Act researchers to compare data on Twitter behavior versus actual behavior?
I am unaware of public agencies that are voluntarily asking for college students’ Twitter handles. However, I think it could be very useful for public health or local law enforcement agencies to use Twitter to monitor and predict health outcomes and crime rates. Government agencies often collect statics on things such as crime rates or health-related outcomes at the county level. In our previous research, we used public health data and combined it with geo-tagged Twitter data to predict outbreaks of HIV. Other researchers have used similar methods to predict obesity, heart disease, and physical activity. I think researchers in this area will be able to increase their predictive accuracy as they discover new methods.
Are there any tools to automate tweets from college students, so that they are not dependent on free will/cognition to manually type and submit text? Is there a way to tie their “sentiments” closer to their stress?
There are current many “Tweet bots” out there that can assist users to tweet about relevant news articles and current events. However, I am unaware of any tools that can link computer-generated tweets to reflect a person’s emotions. I think predicting stress using a single variable such as “sentiment” can be limiting. Using multiple predicting variables and data sources, such as daily steps, heart rate, and sleep pattern, can increase overall model accuracy.
Is Twitter a place where college students turn for prayer?
There are many people who turn to Twitter for prayer. As a matter of fact, Pope Francis uses Twitter to take prayer requests. Twitter can enable individuals to express their religion and receive support, but I think it’s important to protect individuals from things such as hate speech and cyberbullying so that they can express themselves in a safe environment.