Microsoft and NYPD team up to fight crime

The current hit TV show Person of Interest centers on a person who has created a computer system that can analyze information taken from tons of surveillance cameras. The computer system then tries to predict if crimes are going to occur. Now it looks like Microsoft has helped to create a technology service that has some similarities with the fictional computer in the TV show.

A new press release from the New York City government's office has announced that the New York Police Department has teamed up with Microsoft to launch what they are calling the Domain Awareness System. The press release states that it " ... aggregates and analyzes existing public safety data streams in real time, providing NYPD investigators and analysts with a comprehensive view of potential threats and criminal activity. "

The press release says that the Domain Awareness system can be used to give fast alerts for suspicious packages and vehicles using products like smart cameras and license plate readers. It can also be used to track pattern in criminal activities, find out where a car associated with a suspect has been in the city and more. Microsoft is planning to sell this new law enforcement technology service to other cities; the NYPD stands to get 30 percent of all sales of the Domain Awareness System.

Source: NYC.gov press release

Report a problem with article
Previous Story

Steam to challenge Windows Store by offering non-gaming software

Next Story

Some BUILD 2012 events will be shown online

15 Comments

Commenting is disabled on this article.

Just remember, 'slippery slope' is a logical fallacy, despite society deciding somewhere along the way that it's a valid concern.

not surprising

i wonder if the data it generates is going to be valid evidence in a court of law ?
If so that might set a lot of precedences which concerns me.

Pattern-based crime prevention can work, believe it or not. There are always people who kneejerk against this stuff because they think the one person out of a thousand wrongly scanned (who will NOT be charged with a crime) justifies letting the 999 out of a thousand off the hook, and generally oppose advancement of crime fighting technology altogether.

This comes from a fundamental belief by many, many people that the system cannot be 'improved' except by being 'dismantled' and 'rebuilt'.

There's also an inherent hatred for statistics and feeling reduced to 'numbers' that can be predicted. We cling tightly to a need to feel unpredictable, individual, and fully capable of free will. Feeling as if MATH can read our actions raised hackles and makes people feel like something deeply personal is being threatened.

But, frankly, statistics are accurate more often than inaccurate (by their nature) as long as they're used properly. To this day, credit card companies are following the two-tanks-and-a-pair-of-shoes rule, and it works.

Joshie said,
Pattern-based crime prevention can work, believe it or not. There are always people who kneejerk against this stuff because they think the one person out of a thousand wrongly scanned (who will NOT be charged with a crime) justifies letting the 999 out of a thousand off the hook, and generally oppose advancement of crime fighting technology altogether.

This comes from a fundamental belief by many, many people that the system cannot be 'improved' except by being 'dismantled' and 'rebuilt'.

There's also an inherent hatred for statistics and feeling reduced to 'numbers' that can be predicted. We cling tightly to a need to feel unpredictable, individual, and fully capable of free will. Feeling as if MATH can read our actions raised hackles and makes people feel like something deeply personal is being threatened.

But, frankly, statistics are accurate more often than inaccurate (by their nature) as long as they're used properly. To this day, credit card companies are following the two-tanks-and-a-pair-of-shoes rule, and it works.

It's true, you take a data set, make a series of measurements of parameters somehow which generate feature descriptors - long vectors uniquely identifying certain values - and perform Principle Component analysis to find trends http://en.wikipedia.org/wiki/Principle_component_analysis. I learned about this in the context of visual search (like Google has) with images.

Now I'm wondering how difficult it would be to make a web crawler that crawls Neowin, and looks at the frequency of words used by members and tries to classify them somehow...maybe I'm just thinking too much!

How much longer before they invent a system that detects behavior patterns associated with criminal activity? Like how you move the way you look and such. Seems like an inevitability.

Leopoldo Dante Rodriguez said,
How much longer before they invent a system that detects behavior patterns associated with criminal activity? Like how you move the way you look and such. Seems like an inevitability.

What? You don't want that? Got something to hide?

/s

Jesus...

GS:win

Glassed Silver said,

What? You don't want that? Got something to hide?

/s

Jesus...

GS:win


Study bell graphs. The wider the scope, the more false positives you get, and you won't ever have a 100% foolproof system with no false positives.

n_K said,

Study bell graphs. The wider the scope, the more false positives you get, and you won't ever have a 100% foolproof system with no false positives.

Thats why its just a tool and it will only make going through all this data allot easier, faster and more effectively.