Big Data is a term that gets used and abused all the time in the press these days. When we think of Big Data we usually conjure up an image of the Central Intelligence Agency. Google being able to crunch huge amounts of information at breakneck speed. Big Data is meant to quantify our lives in unprecedented ways. That will help improve business, drive innovation and lead to breakthroughs in science. It’s also meant to be transformative for law enforcement, giving investigators new tools for crime-fighting and surveillance.
But it turns out Big Data isn’t all unicorns and rainbows as evidenced by a recent conference hosted by the Massachusetts Institute of Technology (MIT) entitled “Big data: The challenges of doing research at scale.” During the conference, several scholars talked about how Big Data can actually get in the way of research. And this problem doesn’t apply to lofty areas like astronomy, but also to down-to-earth things like crime-fighting.
Here are Technological Hang-Ups Holding Back Big Data:
- To better understand the potential pitfalls of Big Data. let’s take a closer look at what one criminal justice researcher had to say about it during the MIT conference. Rachel Green stadt is an associate professor in Drexel University’s department of computer science. Engineering who studies deception detection. What she revealed is that when it comes to certain kinds of investigative tasks. Where investigators are looking for small but important clues in reams of data, then the methodologies used by Big Data analysts can actually get in the way.
- “We’ve made significant progress over 20 years applying machine learning techniques for finding red flags in financial transactions,” said Green stadt at the conference. “The problem with this is that any small pieces of data can look like a red flag if they’re in the wrong context.”
- For instance, let’s say investigators are looking for someone who makes transactions involving child pornography. The transactions might have nothing to do with child porn at all. It could be an innocent individual purchasing travel tickets or making donations to charities. But based on Big Data algorithms, law enforcement would treat every single person. Who has made similar transactions as a potential suspect. And this problem isn’t just theoretical either. We’ve seen law enforcement make arrests based on false positives before.
- Green stadt also said that Big Data is good for finding trends over time.But it doesn’t work well when you need to focus on specific incidents. It would take too long to go through every single data point.
- Of course, Big Data can also be incredibly helpful in certain investigatory tasks. For instance, it makes location tracking easier than ever before. That could come in handy for catching criminals or finding victims of crimes like human trafficking. But when investigators are searching for small but important clues. Within reams of data—in the same way. A detective would inspect an area in hopes of spotting something out-of-place—then Big Data may sometimes get in the way.
So what’s the solution?
- It depends on what kind of information you’re looking for and how far back you’re trying to go, Green stadt said. She added that sometimes Big Data algorithms fail because they’re simply not designed with specific investigatory tasks in mind. So simply scaling up those algorithms isn’t going to solve the problem either.
- In the end, it all comes down to training investigators to use Big Data. Properly and wisely—and that’s going to require a lot of time and effort.
- The next time someone tells you about a groundbreaking new tool that will revolutionize law enforcement simply by crunching data, remember Green stadt’s words: “Any small pieces of data can look like a red flag if they’re in the wrong context.”
- It also means we need more researchers who understand both how technology works as well as human behavior. In other words, our future Sherlock Holmes might be a criminologist with a Ph.D. from MIT who dabbles in machine learning. Rather than a private detective with a blog.
Conclusion:
Data by itself is not enough. We need more than just data; we need context and understanding of what it all means. This is exactly why we need more criminologists and CSI experts who understand the power of technology. Without them, we risk focusing on false positives and ignoring important pieces of evidence. Perhaps one day we will have a machine that can be trained to find clues. Solve cases like its human counterpart, only time will tell if this is possible.