Does increased "chatter" amongst evil doer bad guy terrorist really indicate an eminent attack of some kind? We hear a lot about this in the global media and from national security consultants and surveillance professionals, but is it really true? Well, it turns out it is, and we know this from hindsight. We also have a pretty good idea when such events will take place based on believe it or not religious holidays, weather, solar storms, economic issues, food shortages, and events made into big deals by the media. All this promises to increase communication and "chatter" and that's part of it.
Still, how do you distinguish from natural spikes of communication, and terrorist "chatter" is the million dollar question. Well, they've figured that out as well. Okay so, let's talk about this shall we?
An interesting paper to read is; "Bursty and Hierarchical Structure in Streams," by Jon Kleinberg and how data mining of emails, research papers in a particular domain or news articles can lead us to find significant "bursts of activity" or a flurry of information or activity on a certain topic in a short period of time. Such tools can signal a population's change in attitude towards a candidate, war effort, or political question of the day. In industry it can also signal upheaval from the norm, or in markets a signal that it's going to be a good day for high frequency trading machines as the VIX or volatility index is ready to come alive.
In many regards these tools and related algorithms can also help us sift through the noise and find an increase in chatter from terrorist elements. Such a flurry of activity pre-leads large scale events, or a barrage of little ones. It also works well for automated systems reading emails coming into a doctors office and catching the intensity, negativity, or pro-status on an issue. Which means it also works pretty damn good on social networks for gauging the minds of the masses.
When a ton of red flags go up in a short proxime of time or there is an unusual uptick in activity between questionable global terrorist network characters, the data mining can alert us to upscale our defenses. If there is activity "across time" from stored data of such characters this too can then be looked at more closely for more clues and a deer dig.
If we look at the bursting activity of news types articles we can easily gauge public perception because people tend to carry opinions of their most read news outlets, thus, if I as a writer was preparing a teleprompter speech for a politician, I'd know exactly where to focus and understanding that "he or they" who control the media, or social network viral topics control the minds of the people. See that point? Data mining thus, in this regard is a no-brainer as well.
Corporations often use such system to data mine text in sentences to help cut down on the large pile of applicants, freeing up time for the Human Resource staff, while ensuring that the most qualified mandates are called in for a second review, the first just being a fancy data mining text algorithm.
Using natural language processing software it turns out we can find out all sorts of things, and when we have flurries of inter-terrorist communication of a specific type, well, we know an attack is eminent, then the question is how to get the pertinent information to preempt. Please consider all this and think on it.