Via Modern Mythology:
The movie Bully is hitting theaters on Friday, and it is making quite a stir.
The issue of bullying in schools has taken a surprisingly long time to reach a mainstream tipping point, considering its link to many of the school shootings that have destroyed countless people’s lives, not to mention its presence in many of psychological makeups as adults. Some people are challenging its narrative as being too simplistic, overlooking all of the other ways in which bullying occurs in our society.
Be that as it may, it is here now. However, the public discussion on the topic has just begun.
That may come as a surprise to some. We all think we can recognize bullying, and certainly sometimes it is obvious: “I’m going to beat you until you can’t walk.”
Few would question that is an instance of bullying. But what about “I’m against abortion 100% … but I hate to tell you what I think about these people ever being born”?
Or, “I’m not homophobic but I don’t think being gay is right. But I know some of the SWEETEST gay people. Dang.”
So much of our understanding of the sentiment behind language comes from inflection, context, and a variety of other cues that are stripped out of communication when it is just text.
Or are they? Plenty of bullying, harassment, or outright hate-speech occurs on Facebook, on Twitter, or in text messages. The sheer shrill volume of hate on the internet makes it seem that early detection of the sentiment behind such statements would be a valuable tool in the public discussion of bullying that is finally underway, if not a method to combat it directly.
There is, in fact, such a tool. There is a great deal of research being done in this direction in many domains, but the Beacon Initiative is putting it to use to track down bullying. What is amazing is that in their on-going beta testing, Beacon’s software has detected all of the previous statements as instances of bullying. To do so, the software had to learn, and to go a great deal beyond simple keyword analysis.
It does this through sentiment analysis. Without getting too technical, your feelings can be “read” through the language that you use.
Of course, sentiment analysis has its limits. The software essentially has Aspergers; it cannot determine when someone is joking or sense inflection, though the case samples above demonstrate that it is starting to learn shades of sarcasm. With this in mind, it would seem best applied to more formal social contexts where certain kinds of joking are generally considered questionable.
The deeper question raised by this software is what to do when such language has been flagged. There can be a fine line between awareness and censorship, which depends entirely on the administrative practices of the organization implementing such technology.
On the Beacon Initiative website, their position on this seems clear enough:
Manipulative speech is most effective on people who don’t know the mechanics behind how misinformation can appear truthful or harmless.
Learning to spot deceptive tactics intended to covertly bully, harass into silence, subtly misinform or deceptively legitimize hateful speech is the best way to empower internet citizens to stand up for justice in real time, and to recognize misinformation when it appears. … Everyone has the power to end hate speech and help end the spread of violent hate crimes where they start — in the mind.
There is a cognitive process that occurs when an individual is made aware of the potential effect of their language. It adds a new layer of self-critique in our own thought process.
This doesn’t reduce the potential for sentiment analysis to be used toward purposes that are less than good. Some may argue, for instance, that the work already done by the NSA toward this end – using sentiment analysis to, among other things, attempt to predict the future is such an example.
Whatever your stance on the issue of bullying, or on campaigns aimed at dealing with it in one way or another, one thing is for sure: bullying is real, it is complicated, and it isn’t just something that some people should “get over.” And this technology exists, right now, and it is going to be used. Who do you think should be using it, and how?