By Joe Wright
A whole host of reasons are being given to justify the increased surveillance and data mining of social media: terrorism, suicide prevention, precrime detection of violence and other illegal acts … and now teenage alcohol abuse.
Pattern recognition is at the heart of data mining for both public organizations and private companies. The Feds tried to get Congress to sign on to required reporting of “suspicious activity” in social media with the Intelligence Authorization Act of 2016, but that provision was ultimately rejected. However, the voluntary two-way data stream inherently provided by social media is proving to be irresistible for monitoring “bad behavior.”
A number of university press releases have revealed what type of research is ongoing, always to counter some perceived threat to society at large.
After news of Ebola began terrorizing the planet, Virginia University announced a program called ChatterGrabber which was developed to detect public health risks. However, it was apparently designed to do much more than that, and only used the dread of contracting Ebola as the ultimate way to get people sympathetic to the idea:
ChatterGrabber has also been used to monitor tickborne diseases, such as Lyme disease, public sentiment involving vaccines, and gun violence and terrorism, serving as an early warning system for public health officials through suspicious tweets or conversations. (emphasis added) [Source]
MIT followed suit shortly thereafter by announcing that Twitter would permit access to all of its tweets beginning in 2006 in a search for how “social movements” take place. Reddit was also mentioned as slated for the same.
But it was Facebook that caused the most outrage thus far with its “Truthy” experiment that went directly into harvesting political data to manipulate the spread of information in a mass psychological experiment on its users.
With that backdrop, any seemingly benign protective measure that uses the same technology mentioned above must be called into question.
The University of Rochester’s full press release is posted below, emphasis added.
Instagram could offer a novel way of monitoring the drinking habits of teenagers.
Using photos and text from Instagram, a team of researchers from the University of Rochester has shown that this data can not only expose patterns of underage drinking more cheaply and faster than conventional surveys, but also find new patterns, such as what alcohol brands or types are favored by different demographic groups. The researchers say they hope exposing these patterns could help develop effective intervention.
Instagram is very popular among teenagers and it offers large amounts of information about this target population in the form of photos and text. As Jiebo Luo, professor of computer science at the University of Rochester, and his colleagues describe in a new paper, underage drinkers “are willing to share their alcohol consumption experience” in social media. Studying the social media behavior of this group allows the researchers to observe it passively in an “undisturbed state.”
They are presenting their work this week at the 2015 IEEE International Conference on Big Data in Santa Clara, California.
An example of the disadvantages of traditional methods for monitoring underage alcohol consumption is that teenagers might not be honest when they respond to an administered survey about alcohol use, e.g., the “Monitoring the Future” survey by the federal government. Also, those that choose to respond to such a survey might not be a representative sample and the sample size might be too small to draw conclusions.
Instagram does not offer a way of selecting users by age, but the research team was able to select users that fit the profile they were looking for by applying computer vision techniques. Luo and his team have been pioneering techniques that teach computers to extract information from images on the Internet (some examples are found here and here), something that is much more complex than just extracting information from text. They were able to use computers to analyze the profile faces of Instagram users to get sufficiently accurate guesses for their age, gender, and race.
Having selected a group of underage users to study, the researchers monitored drinking related activities via their Instagram photos by analyzing the social media tags associated with these photos using a constructed Internet slang dictionary and also any alcohol brands the users follow.
In their study, the researchers found that underage alcohol consumption, like with adults, happens more on weekends and holidays and at the end of the day. There also wasn’t a strong bias toward one gender for alcohol consumption – it matched the gender ratio of Instagram users.
The researchers did find that different alcohol brands are followed in varying degrees by teenagers, and that different genders follow different brands. The researchers highlight that this could point out brands that are attracting younger audiences in social media, information that could be useful to people working with underage drinkers.
“There are several ways we can go about doing that,” said Luo. “We can keep government agencies or schools better informed and help them design interventions. We could also use social media to incorporate targeted intervention and to measure the effect of any intervention. And perhaps other things we haven’t thought about.”
The researchers acknowledge, however, that research like this could also be used by brands to target their products to those users most likely to follow them.
Luo explained that an important next step is to check the results of their approach with surveys, to ensure their methodology is robust before applying it to extract even more information from Instagram. They hope to collaborate with people working on addressing other youth problems, such as tobacco, drugs, teen pregnancy, stress or depression.
“This new method could be a useful complement to more traditional methods of measuring youth drinking,” said Elizabeth Handley, clinical psychologist and research associate at the University’s Mount Hope Family Center. “It could provide important new insights into the contexts of youth drinking and be a valuable tool for evaluating the effectiveness of school or community-based preventive interventions.”
Luo’s coauthors on the paper are Ran Pang, Agustin Baretto, and Professor Henry Kautz, all from the University of Rochester.
The repeated use of the term “intervention” is what should sound alarms the loudest; not only are they collecting data, but they intend to take action based upon what they find. Yes, we are largely dealing with minors who have not been granted the complete rights of adults, but this same language has been used in predictive mental health algorithms designed to do the exact same things for all ages. And, as we should have learned by now, loose definitions and loose interpretations have led to repeated abuses in a world that seemingly has gone full paranoid about all potential threats.
Do you feel that this technology is a legitimate solution for this specific issue, or is it destined to take us one more step down the slippery slope?
Joe Wright’s articles can be found on ActivistPost.com