Browse Prior Art Database

System for Detecting Probable Criminal Activity Based on Past "Criminal Transactions" Disclosure Number: IPCOM000238933D
Publication Date: 2014-Sep-25
Document File: 2 page(s) / 44K

Publishing Venue

The Prior Art Database


This article describes a system for alerting police officers to crime patterns in near real time as they are emerging. The system has two distinct parts, the first one is the pre-compilation of "criminal preferences" and the second is the near real time alerting mechanism. 1. Criminal Preferences Pre-Compilation * Data is aggregated from disparate data sources and consolidated using a previously patented rules engine, * An algorithm is run on this data to create "criminal preferences" for types of crime, MO's, etc. 2. Near Real Time Alerting * As 911 calls come in to a call center, the data stream (either voice or manually entered) is compared against the criminal preferences generated in step 1 above. * If there are statistically significant matches between the incoming data stream and the pre-compiled preferences, alerts are generated for officers in the field and command staff to enhance situational awareness and aid in predicting the eventual outcomes of an event in progress.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 41% of the total text.

Page 01 of 2

System for Detecting Probable Criminal Activity Based on Past "

Transactions "

Law enforcement officers have many years of experience observing crime trends that are local to their individual beats and they use that knowledge to be pro-active in predicting and thus preventing future crimes. For example, a former detective in Tucson once observed that in his experience , areas where a spike in "prowler" or "peeping Tom" activity occurs are at a higher risk for home invasion and rape in the near future. He also relayed a story where a spike of these prowler calls occurred but there were no obvious leads to follow. After investigating these occurrences for a few days, the detective noticed that there were matching bicycle tracks near the incidents. He then went into his file system and searched for Field Interviews of people on bicycles in that area. After searching these records and checking the background of the person being followed, he was able to identify that this was indeed the suspect involved in the prowler activity and that he had a record of prior rapes in other areas .

The invention described below takes this multi-day manual process, automates it and compresses it to minutes by combining data mining techniques, such as event co-occurrence, with the unique capabilities built into the IBM i2 COPLINK platform.

An exhaustive study on other predictive policing solutions is referenced here :

Drawbacks of other systems using predictive policing include
1. Analyzing the wrong data (e.g. arrest data)

This proposed solution is different in that it trains on historical data (consolidated RDBMS records) which is mapped to a different type of incoming data (911 calls).

2. "So far, predictions have mostly been made about people who have already had contact with the justice system-such as convicted criminals. "
--Reference: wrongdoers-dont-even-think-about-it. In the example, those people are included in the analysis, but their associates who may not be convicted criminals are also included . This is a key difference between this invention and those known at this time.

With any predictive policing solution, one of the drawbacks is the accuracy of the predictions and the liability (if any) of using these predictions in the course of an arrest. In the case of all COPLINK software, we constantly put forth the message to the customers and the public that the software are only tools to be used by trained law enforcement professionals in the course of their sworn duties . We do not purport to supplant a law enforcement officer's own experience and judgment with any of the software. This is in line with the "comprehensive business process" documented on page xvii of the Rand paper mentioned above.

Further, we take measure...