Apparently Hitachi thinks we can.
Remember the experimental “Precrime” crime-prediction department headed up by Tom Cruise as Chief John Anderton in the 2002 Move Minority Report?
Well, Hitachi announced this week that it has developed a powerful and robust new technology called ‘Hitachi Visualization Predictive Crime Analytics’ that has the ability to pinpoint when and where a crime will occur
By gathering data from a fire hose of multiple real-time sources, including:
- Historical crime statistics
- Proximity to schools and subways
- Gunshot sensors
- License plate readers
- Public transit maps
- Weather reports
- and of course Social Media Conversations
the system then uses machine learning to find patterns that humans can’t see with the goal of making our cities safer places to live.
“A human just can’t handle when you get to the tens or hundreds of variables that could impact crime,” Darrin Lipscomb, an executive in Hitachi’s Public Safety and Visualization division, told Fast Company.
It’s mind blowing how the Hitachi Visualization Predictive Crime Analytics system can specify potential crime scenes down to a relatively small 200-square-meter spot, and it assigns relative threat levels to every situation!
Traditionally, police investigators build crime-prediction models based on their experience with certain variables, such as the location of schools or slang words for drugs that pop up on Twitter. They assign a weight to each variable based on how important it seems to be. This is now streamlined with Hitachi’s system which doesn’t require a human to figure out what variables matter and how much.
“You just feed those data sets,” says Mark Jules, Hitachi’s Public Safety and Visualization division
“And it decides, over a couple of weeks, if there a correlation.”
The main benefit of the system is to create rich actionable insights and threat level predictions that can accurately forecast both when and where crimes are most likely to happen and whether or not additional resources are likely to be needed in the area.
Social Media In Prediction
Social media is often used in solving crimes: from soliciting crime tips, to identifying people and their locations, gathering evidence, and notifying the public, law enforcement has been using social media for some time.
This also isn’t the first time social media has been used in prediction. Twitter has previously been used to predict flu epidemics and even predict the stock market!
Traditionally, police investigators build crime-prediction models based on their experience with certain variables, such as the location of schools or slang words for drugs that pop up on Twitter. They assign a weight to each variable based on how important it seems to be. This is now streamlined with Hitachi’s system which doesn’t require a human to figure out what variables matter and how much.
“You just feed those data sets,” says Mark Jules, Hitachi’s Public Safety and Visualization division “And it decides, over a couple of weeks, if there a correlation.”
The main benefit of the system is to create rich actionable insights and threat level predictions that can accurately forecast both when and where crimes are most likely to happen and whether or not additional resources are likely to be needed in the area.
Social Media In Prediction
Social media is often used in solving crimes: from soliciting crime tips, to identifying people and their locations, gathering evidence, and notifying the public, law enforcement has been using social media for some time.
This also isn’t the first time social media has been used in prediction. Twitter has previously been used to predict flu epidemics and even predict the stock market!
As far as predicting crime goes, social media plays a big role and it can even improve accuracy by 15%.
Hitachi uses natural language processing: the ability of a computer to take in and understand colloquial text or speech.
The system can sort through every single tweet tagged to a specific geography to find the most significant words that indicate what’s happening.
Gangs, for instance, often use different keywords to meet up or coordinate activity. While we may not know what that keyword is, the software can identify anything that’s abnormal, such as someone using an off-topic word, or using it in a very tight density or proximity, and then assign it a bigger weight in the results.
One thing social media indicates is tension between neighborhoods that could turn violent. “We were talking to [Washington] D.C., and they said, our biggest cause and effect is what neighborhood you’re closest to,” says Lipscomb. “There are these neighborhood rivalries going on in D.C.”
“Normally”, said Jules, “police wouldn’t realize the correlation between neighborhood tension flare-ups and crime until months later.”
Does It Really Work?
The company plans to put the system into a trial run at police departments in a handful of unspecified cities starting in October, but it raises two major concerns:
1) How accurate will the system be?
and
2) What if the system unfairly profiles and targets innocent people as criminals?
To address accuracy, Hitachi has promised to make all of its trial results public, and allow people to decide its effectiveness for themselves.
As for the second issue, Hitachi execs say the system may actually reduce the amount of biased police profiling, since it will equip officers with enough thorough information that they won’t need to act on mere suspicions.
What are your thoughts on a system like this? Is it useful? Or is our privacy being invaded more and more? Comment below!
