Army engineers want to use Facebook and Twitter on the front lines.
Believe it and retweet it. The engineers say commanders could benefit from having a finer understanding of the social media landscape in a conflict zone.
“Social media is a new channel that offers massive amounts of information,” said Reginald Hobbs, branch chief of the multilingual computing and analytics groups at the U.S. Army Research Laboratory. “In the past people would be consumers of information. Social media makes people producers of information, and the Army can leverage that.”
In their search to turn social information into a tactical tool, Army scientists recently wrote a paper on “social sensing” accepted for publication by the IEEE Computer Society. Still in its infancy, the science here faces some challenges: The volume of information is vast, and it’s largely unstructured. But researchers say they are making headway.
Action and reaction
To understand the value of social information, it’s helpful to keep in mind the breadth of the military role these days. In addition to warfighting, troops may find themselves on the ground during civil emergencies or in situations of perilous stability. In such cases, it helps to have the pulse of the population.
“You want understand the people that are being affected, the mood of the civilian population and the impact our operations might have on the mood,” said Lance Kaplan, a team leader in the network sensing and fusion branch at ARL. “Every action that the Army takes can possibly have a collateral effect on the civilian population, and social media gives us the means to understand that effect.”
It isn’t hard to know what people are thinking: Social media produces a veritable firehose of thoughts, ideas and opinions. The challenge here lies in sifting out the content from the noise, understanding what is true, and what is relevant.
“You have a reliability issue,” Kaplan said. “You are dealing with unknown sources, unknown credibility, so you have to find a way to determine the reliability of those sources.”
To that end, researchers are developing automated tools to seek out correlations. If multiple independent, unrelated sources are saying the same thing, there’s a greater likelihood that it is true.
By using such tools, “we come up with an estimate of the social network influence, who is connected to whom and who is influencing whom,” Kaplan said. “That allows you to untangle the thread. It gives you a probability as to the independence of the information, which helps you to estimate the credibility of the information.”
Emerging tools also scan social sentiment for signs of polarization. “A lot of social media data is opinion,” Hobbs said. By paring away the most extreme views, “it helps you to separate out opinion from fact.”
The scientists say that in order to fully exploit the potential of social sentiment, it’s important not just to understand who is speaking, but to dig into the ways in which they convey their messages.
“We can look at the particular word choices to determine whether the message is trying to use persuasion or influence, as opposed just reporting something. That can give you hints as to whether something is a piece of propaganda as opposed to something with real information,” Kaplan said.
This kind of information can have direct battlefield implications.
Hobbs offers an urban analogy. Say you’re in the city and traffic suddenly comes to a halt. There could be any number of reasons for this. “If you had access to local social media you might determine the local sports team had just won a championship. Maybe that changes your plan,” he said.
The same holds true on the battlefield, where a sudden shift of enemy position or behavior can occur for any of a dozen reasons. “If we can identify people talking about specific entities or specific events, it gives us a deeper understanding of that situation,” he said.
The U.S. Army isn’t alone in pursuing such capabilities. The Jerusalem Post reported recently that Israel’s Shin Bet security service had prevented 250 terrorist attacks in 2018 thanks in large measure to its analysis of social media.
Development of these tools is an ongoing process. As they get better at understanding the content of social data, scientists say, they will be looking for improved ways to automate the analysis.