The
Need for Information-led Policing
SPSS'
Robert Martin explores the latest developments in the use of data across
UK and European police forces.
Controversy often surrounds Home Secretary Jacqui Smith when speaking
of the UK's police force. However, she has stood by a decision to allocate
£50 million to the police to spend on handheld devices such as PDAs.
The PDAs will allow officers more time in solving crime and helping the
local community, and less time at a desk in the station.
This
is all in addition to funding last year for mobile computer devices that
allow fingerprints to be taken on the spot, give officers a chance to
cross-reference a national database of mug shots, and enable information
to be sent back to a central database.
However, the Government's initiatives to increase the level of technology
available to the public sector may not be just a case of spending more
- but also spending more wisely.
It's important not just to acquire data from incidents, the public and
police officers on the beat; but also to be able to quickly assimilate
and interpret that data and allow the searching of information obtained
both quickly and easily.
Many police forces already use techniques such as Predictive Analytics
to look at transient patterns in data and then use link analysis to highlight
a trend. However, the full potential for this technology when working
with information surrounding criminal behaviour is often left untapped.
A full understanding of Predictive Analytics allows the police to do far
more than just counting crime and move onto anticipating, preventing and
perhaps even allowing the community to respond effectively to any threat
- such as a the recent rise in knife crime among young people.
Predicting the outcome
Counting crime and surveying police officers and the victims to determine
how they were dealt with is an important application, but is only one
use of Predictive Analytics. It is known that criminal behaviour tends
to be relatively repetitive. By analysing the data, we can be much more
proactive in foreseeing and preventing crime than at present.
The importance of analysis becomes apparent when you consider that data
which is seemingly irrelevant to one case can often help to solve another.
Based on the technique of profiling, even things like seasonality and
the weather conditions at the time of an incident may become important.
Criminals generally work within their comfort zone and by using Predictive
Analytics, police can often link the modus operandi of one crime to another.
Data collection
Police forces have become very good at collecting data. However, it is
the use of this data where some forces still need to gain ground, especially
with the increasing use of electronic records management systems.
Witness reports often represent unstructured text and so Text Mining,
which can spot patterns in free text to link witness statements together,
is essential in solving crimes that can sometimes span back fifty years.
Often, reports taken by different officers can contain discrepancies -
police officers abbreviate their words in reports and so statements from
different officers often don't match. Also, in some communities,
English is not always the first language. Therefore, only by data sharing
and mining can intelligence be gained.
At present, many UK police forces use double key entry when compiling
reports. This means that when a policeman fills out information for a
report it is called through to a crime reporter who inputs the data onto
a central information system.
The double entry of information can often lead to errors and is labour
intensive. Mobile data collection, used by some police forces internationally,
allows the data to be collected at source, often the scene of a crime
or incident, and then captured and analysed straight away.
It is important then that data is controlled at the point of collection.
If you control the access and type of information collected it is much
quicker and easier to analyse.
The analysis
Once data is collected in a meaningful way, the next step is analysis.
This is one area hampered by a current lack of standardisation across
UK police forces. There is a general consensus that separate approaches
by the 43 forces within the UK to analysing data can slow down detection.
When analysing any data it is important to confirm what data is already
held and what is known about crimes, as well as to discover new relationships
in the data.
The speed and accuracy of analysis becomes even more important when we
consider the average time taken by police to solve major crimes. This
has been increasing over the last 10 years - not due to poor detection
techniques - but more because crimes are often not planned locally any
longer. Generally, crimes are becoming more complicated and can now be
associated with international crime syndicates and terrorists.
Predictive Modelling
After analysing the data, it is possible to filter it and find the key
elements that can provide pointers to possible terrorism or crime.
The police and intelligence services are under enormous pressure to act
faster as well as be more economical. Therefore data mining tools can
look at matches in the statistics very rapidly to sift textual data allowing
real time analysis.
For a large organisation, the importance of sharing data is paramount.
It is vital to look at all transcripts and then find any links quickly.
The Dutch national police agency (KLPD) use a 'digital washing machine'
which uses data cleaning to extract information from computer files and
filters out only the information that is important. The information is
then converted into text files before being able to conduct text mining
to show existing and potential relationships. The system allows the Dutch
police to not only check sub-cases and their notes, but also Internet
data that can amount to terabytes in size.
Public v. private
The private sector has led the field in information technology and can
provide enormous technology transfer advantages to the public sector.
Jacqui Smith's announcement of PSA 23 (Public Service Agreement) - a charter
designed to make communities safer - will mean more emphasis on the police
to not only detect, but also prevent, crime so we can identify individuals
before they become fully-fledged criminals.
US and European police forces already use Predictive Analytics for these
purposes. 224 of forces worldwide use technology from SPSS on a tactical,
as well as the detection, level. They forecast hotspots that may arise
at the beat level and use the predictive technology to give multiple indicators
on what is going to happen. This can be combined with incident statistics
from the past to determine the most likely outcome.The longer the time
between the incident and the actual analysis of and reporting on the evidence,
with results in a useable form, the less valuable it is. It is therefore
increasingly essential that data plays a more central role in the fight
against crime.
