See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235322994

Police crime recording and investigation

systems A users view

Article in Policing An International Journal of Police Strategies and Management March 2001
DOI: 10.1108/13639510110382287

CITATIONS READS

24 1,015

2 authors, including:

Richard Adderley
University of Leicester
27 PUBLICATIONS 207 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Valcri View project

Mosaic View project

All content following this page was uploaded by Richard Adderley on 30 July 2014.

The user has requested enhancement of the downloaded file.

 Police Crime Recording and Investigation Systems:

A Users View

INTRODUCTION

Recent years have seen an increasing use of computers within Police

Forces, not just in general administration (e.g. payroll, personnel, e-mail, general

office suites etc.), but also as a tool to assist the core business activity of

policing (preventing and solving) crime. Increasing crime rates and the (greater

awareness of modern criminals) puts a strain on existing methods and

techniques. There is therefore a clear need for the application of advanced

software technologies in order to make the best use of limited resources.

The objective of this paper is to provide an insight into the role software

systems are playing in Police Forces, primarily in the United Kingdom but also

in North America and Europe. Many of these systems are in-house developments

by individual Police Forces or commercial products, hence few details can be

found in the academic literature. Some of the information contained within this

paper was gathered by word of mouth from the developers. As a consequence

many of the details of specific systems are sketchy. It is hoped however, that this

paper will be of interest not only to those working on related systems but also to

those with a general interest in how advanced software technologies are being

applied to a challenging application

area.

Print Date: 19/10/06

 THE TASK

Crime occurs in a variety of forms. Police informally categorise crime as

being either major or volume.

Major crime consists of the high profile crimes such as murder, armed

robbery and non-date rape. These crimes can either be one-offs or serial. In the

case of serial crimes it is relatively easy to link crimes together due to clear

similarities in terms of modus operandi or descriptions of offenders. This

linking is possible due to the comparatively low volume of such crimes. Major

crimes usually have a team of detectives allocated to conduct the investigation.

In contrast volume crimes such as burglary and shoplifting are far more

prevalent. They are usually serial in nature as offenders go on to commit many

such crimes. Property crimes, such as domestic burglary offences, committed

by different individuals are highly similar and it is rare to have a description of

the offenders. This makes it difficult to link crimes together. Also a single

Detective Officer may have up to 40 different volume crimes to investigate at

any point in time. Within the West Midlands Police area, in the last decade,

total crime has risen by 15.1% whereas police establishment has only risen

7.7%.

With a view to satisfying this demand Police Forces around the world

employ specialist Crime Analysts, people who have specialist training in a

variety of disciplines including investigation techniques, criminal psychology

and information technology. It is their task to assist investigating Officers by

analysing crime trends and patterns, identifying links between crimes and

producing packages which targets an individual or group of offenders linking

them to a series of crimes.

2
 Most, if not all, of current systems both manual and computerised revolve

around the investigation of crimes already committed. They are, therefore,

reactive. The majority of UK Forces use different types of relational database

management systems (RDBMS) for recording and subsequent analysis of

crime. Standard or interactive queries are written to produce patterns of crime,

offending and various statistics.

SPECIAL PURPOSE SOFTWARE TOOLS

The development of software for dedicated tasks such as fingerprint

matching, scene of crime reconstruction and vehicle number plate recognition

have been areas of active research for a number of years. These systems can be

used as tools for major or volume crime but their dedicated nature (i.e. a well

defined objective) lies outside the scope of the current review into crime

analysis. They are also covered well in the academic literature and the

interested reader should consult the cited literature for details.

COMPUTERISED CRIME SYSTEMS FOR DEALING WITH

MAJOR CRIMES

The investigation of a major crime generates a large volume of data and

information related to the particular crime, or series of crime. This data can

overwhelm any single individual. Partitioning the data between team members

may not help due to the partition obscuring possible links. This problem has

motivated a variety of approaches to use the power of computers to assist in the

management of this voluminous data.

Home Office Large Major Enquiry System (HOLMES)

All 43 UK Police Forces use the current HOLMES system that is

approximately 12 years old. There are 5 variant main systems produced by

3
Bull, ICL, CGS UK Ltd., Unisys and MDIS each using its own proprietary

database, programming and query language. Information cannot easily be

transferred between systems, in fact, there is only a limited linking facility

between systems from the same manufacturer.

The West Midlands Police use the Bull system that has 54 different files,

which are accessed by the various programs written in a proprietary language.

Embedded within the system is the Status text processing program. HOLMES

uses a Bull utility, Transaction Processing System (TPS), to extract text from

the files and compile the Status database. To search for red Vauxhall the

system will instigate a procedure, which is transparent to the user, by firstly

searching the vehicle file and then the Status database.

Within the present HOLMES systems all data is limited to a single

incident, no searching can occur across incidents thereby imposing obvious

limitations.

HOLMES2 began its testing phase at the end of 1997 offering many

advantages over the existing system. It uses a Dynamic Reasoning Engine (DRE)

within a RDBMS in a Microsoft Windows environment that has been developed

by Cambridge Neurodynamics Limited, England [1]. The engine is capable of

searching all incidents on the system with the facility of using natural

language searches. The results are a list of documents that are possibly linked to

or have a relationship with previously unconnected documents and a list of

possible keywords is produced which may be used in future searches. The

engine may be used in automatic mode, manual mode or a combination of these

modes thereby allowing the enquirer full control of the AI used.

4
Compatibly between different Police Forces is ensured as the system was

designed and built by a single contractor.

Interpol

Interpol has been the centre of police co-operation for its 177 member

states for almost 75 years. Its computer system centralises data from members'

National Central Bureaux (NCBs) which amount to 100 000 new documents

annually. These are stored and archived in the main database in Lyon, France.

Data is enriched and annotated and a selection is made available to NCBs

through an Interpol designed Automated Search Facility. It is now moving

from its mainframe-based system to IBM RS/6000s running relational database

technology from Oracle [2].

The project was launched in 1996 with the objective of integrating the

three diverse systems of the core criminal database, the Electronic Archive and

the Automated Search Facility. It was expected to roll out the system in late

1997 when there will be two copies of the integrated database: a master copy

for standard search facilities and a replicated copy supporting the use of more

complex crime analysis tools. The visualisation tool, i2 (see below for software

description) will be used to analyse complex searches through telephone or

bank account numbers. Free text searching across the database will retrieve

information that is often received out of context and the ability to store very high

resolution colour images will assist in the tracking of stolen works of art, credit

cards etc. It is now possible to provide an image of a stolen or counterfeit credit

card to the central system and it being made available to all member states

worldwide within hours.

5
 Violent Criminal Apprehension System. (VICAP)

The Federal Bureau of Investigation developed the VICAP system that

uses pattern analysis to monitor case activity enabling the investigator(s) to

trace the travelling criminal who is actively involved in major violent crime. The

types of cases examined by this system include: (i) solved or unsolved murders

(or attempted murders) that are apparently random, motiveless or sexually

oriented; or are known or suspected to be part of a series (ii) missing persons

where there is suspicion of foul play and (iii) unidentified dead bodies where

murder is suspected.

When a new case is entered, the system simultaneously compares and

contrasts 188 variables relating to MO categories against those held in the

database. The output is a listing of the ten cases that most resemble the one being

entered. The VICAP system has case management administration facilities,

investigation support (time lines, investigative matrix etc.) and facilitates crime

investigation training.

Criminal Geographic Targeting (CGT)

During 1995 the Vancouver Police Department Police Department

introduced the CGT system which was developed by one of their Officers, Dr.

Kim Rossmo [3]. This system investigates the probable spatial behaviour of an

offender by examining information which can be found at known crime site

locations such as encounter/apprehension sites; murder scenes; body/property

dump sites.

The system reverses the approach of environmental criminologists who

use sociological information such as the offender's background, peer influences

criminal careers etc. to describe, understand and control criminal events. The

6
relationship between an offender's home and the crimes he or she commits is

the underlying theme of their work. Rossmo reverses this reasoning to predict

the most probable location of a criminal's residence by using the locations of a

series of crimes. The result is a three-dimensional plot, the height representing

the relative probability that a given point is the residence or workplace of the

offender that is overlaid onto a local map.

The system was initially validated on several solved crimes locating the

offender's residence in the top 5% or potential locations. Subsequently the

system has had a number of successes [4], in one notorious case the then

current investigative systems suggested that the killer could have lived within a

2,000 square kilometre area, Rossmo's software indicated that the search

should be restricted to a 10 square kilometre area. The offender, who lived within

that restricted area, was subsequently captured and convicted. It was also used

in a case where a 16-year-old girl was murdered and another left for dead. The

software suggested a search area of 1.5 square kilometres from a possible area

of 26 square kilometres. The offender who lived within that search area was

arrested and charged with the offences.

This system is now commercially produced under the name ORION [5].

COMPUTERISED CRIME SYSTEMS FOR DEALING WITH

VOLUME CRIMES

Most crime is not major. Individual crimes are of a less serious nature to

society as a whole (although still serious to the victims). The shear volume of

such crimes creates different types of problems for the investigator. The high

degree of similarity between crimes and the lack of resources for detailed

investigation of any one particular crime creates problems in both solving a

7
crime and preventing similar future crimes. The challenge for advanced software

developers is to apply modern pattern matching and visualisation techniques to

alleviate the problem.

Crime Report Information System (CRIS)

In the late 1980's the London Metropolitan Police conceived CRIS [6] as

a computerised system to replace the paper based crime reporting system with

a database covering the entire Metropolitan Police area. Various solutions were

tried but it was only in August 1994 when a contract was agreed with EDS Ltd.

based on new high performance computers and data networks that the vision

could be realised.

The information is stored in an Ingress database distributed across two VAX

7600 central computers that mirrors all 63 Divisional databases and is updated

in real time [7]. Each Division has both a front-end and back-end processor

connected by X25 network and each Sub-Division is connected to this by a

multiplexer. A Graphical Query Language (GQL) is currently being used to

access information and download into Microsoft Office applications and

MapInfo for geographically plotting 'hot spots'.

Searches undertaken using divisional data are 'held' on the divisional

computer, however, cross-boarder searches are viable and are conducted

centrally. As the data is stored in real time any requests for information will

automatically include the most up to date results. The system has produced

information that has lead to many arrests, the first reported being related to car

crime [8]. Officers used the system to build up a picture of a car crime wave in

Surbitan, Southwest London and found that they followed a pattern of time,

location, method and type of car stolen. The Police established a decoy vehicle

8
and caught the offenders within two hours. This is an illustrative example of how

a computerised criminal investigation software can be used for pro-active

policing.

Crime Pattern Analysis System (CPAS)

The West Midlands Police use CPAS which provides digital maps to display

all or specific types of crime that occur in a defined area. The system will analyse

the data and produce a coloured graphical overlay on the map to show the

density of crime. It also has the facility, using statistical techniques, to

extrapolate total crime figures for a series of dates and predict where crime will

occur 'tomorrow'.

Figure 1. Crime density levels

The prediction is overlaid onto the map. This information can be used to

aid manpower planning and resource allocation. It is not currently used for

solving specific individual crimes or series of crimes committed by the same

offender(s).

Action Information Management (AIM)

During 1997 Suffolk Police piloted a crime mapping system which,

extracts information from their recording systems to display, via Action

9
Information Management software, instances of crime onto an Ordinance Survey

map of the relevant area. The system will plot all/different types of crime and

export summary data into spreadsheets for reporting or further analysis. By

indicating a single crime it is possible to plot all offenders that are associated to

that crime thereby indicating the distance travelled to commit the crime. By

indicating an offender it will plot all crimes associated with that person thereby

indicating the radius of criminal activity including date/time and type of crime

committed.

Figure 2. Crime Buffer Zones

This system has the ability to plot crime corridors by placing all known

addresses (i.e. offenders and known associates) of an offender on the map and

then connecting them by drawing a line along the streets. It is then possible for

the system to create a 'buffer zone' (a polygon) around the street lines at a user

defined distance i.e. 100 meters. All of the crimes within the zone for which

the offender has been charged are then displayed followed by all similar

10
10
outstanding crimes. This will enable detectives to target the offender's activity

or, if already in custody, be ready to further question.

This is illustrated in the diagram below, the black squares representing the

crimes that fall within the zone.

i2

All of the UK, Dutch, Swedish and New Zealand Police Forces, various

Australian Forces and USA legal departments (including the FBI) [9] utilise

the data visualisation suite of software, Analyst Notebook. There are four main

areas within this suite: -

Link Analysis

This type of diagram displays the associations between different entities,

which may be a person, vehicle, organisation, telephone number etc. It will

also show the flow of commodities such as money or property.

Network Analysis

Network diagrams are used on large data sets to extend the link analysis

concepts. This area is mainly used to analyse telephone transactions, account

transfers and Internet traffic. The system will allow optical character reading

(OCR) of information such as telephone bills and automatically displaying the

related links.

Event Sequence

This time line chart reveals how related events unfold. It is particularly

useful in visualising the cause and effect between events and sequence

simulation by automatically arranging events in chronological order and

branching between them.

Transaction Pattern Analysis

11
11
 The chart analyses the significant activity between subjects. It is particularly

useful in visualising repeated activity patterns such as a sequence of telephone

calls indicating a chain of command or a sequence of account transfers indicating

money laundering.

The i2 software was used in the preparation for and during the Euro 96

football tournament [10] where in excess of 7500 potential troublemakers were

expected to arrive in England. Specialist Officers used the software to reveal

unexpected links between apparently unconnected people.

Watson

Watson is an integrated suite of investigative case management and

intelligence analysis software used by several UK and Australian Police

Forces, many USA Police Departments and various UK and USA government

departments [11].

This operational intelligence tool has the ability to track and interpret data

from a wide range of sources by connecting to the source or importing the data,

automatically displaying links. Questions can be formulated without knowing

the data contents: -

1. Ask a question by specifying the objects of your enquiry

2. Receive the answer in a chart

3. Analyse the chart to identify facts or create new lines of enquiry

4. Ask another question

Queries can be saved for later use or applied to different data sets. Using a

'Visualisation by Query' methodology, Watson will produce custom designed

charts, links between data and/or events, hierarchies and transactions. The chart

displays are similar to those in i2. It will also generate activity timelines and

12
12
reports. The linking includes investigating networks of contacts, cashflow

transactions and telephone analysis.

The case management tool, which seamlessly integrates with Watson,

enables Senior Investigating Officers (SIO's) to maintain a structured and

methodological approach to the investigation thereby reducing the amount of

administration. PowerIndexingTM automatically cross-references database

information identifying and tagging records and objects then identifying their

relationship to one another. An auto research function can be applied to

transaction data i.e. telephone calls, cash transactions, which highlights

relationships between, possibly, thousands of records. There are three

management modules within this tool: Action Management, the identification

of relevant lines of enquiry; Task Management, the allocation of specific

tasks to specific people allowing SIO's to monitor the workflow and prioritise

tasks and resource assignments; Document Management - tracking of the

extent that a document was processed, by who and when. It was last updated.

Watson was used in a notorious UK murder case (the Naomi Smith case

[10] ) by linking into HOLMES data producing data links between people,

vehicles, crime scenes etc. could be visually displayed (e.g. question: "was

there any relationship between Mr X and Mr Y"). In this way a large number

of attributes were matched at one time. While detectives were utilising this

information, scientists made a breakthrough in identifying the offender's DNA

that resulted in the undertaking that all males between 15 and 28 years old would

be DNA tested. Watson, together with HOLMES, identified a reduced set of

males to be initially tested and, during the testing process, the offender's DNA

matched one of the first fifteen samples.

13
13
 Suspect Prioritization System (SPS)

Royal Newfoundland Constabulary (RNC) established a Criminal

Behaviour Analysis Unit (CBAU) tasked with providing an offender profiling

service. This was developed based on well documented academic research which

suggested that procedures could be developed which could analyse offender

data thereby provide investigators with a list of suspects in a prioritised

order. However this system goes further by correlating crime behaviour to

offender characteristics [12].

The Information Technology Division of the Constabulary implemented a

database, to record data, linked to a MapInfo Geographical Information system

(GIS) to visualise the crime patterns. In excess of 7000 Offenders criminal

history was entered into the system and continues to be expanded by the entering

of current overnight arrest information. The system generates a report indicating

suspect(s) in ranked order

The system although not yet in general use was used to by the originator of

the system who had property stolen from his car. The crime was profiled by the

system, the top five suspects were investigated and the offender was found to

be the third ranked suspect.

Regional Crime Analysis Project (ReCAP)

This project has been implemented by the University of Virginia in

Charlottesville along with their local Police departments for use by crime

analysts [13]. ReCAP uses a GIS with statistical analysis and machine learning

to examine a corporate database that is used by the City and County Police

Forces as well as the University. This subsequently allows the rule based

system access to relevant information across Police jurisdictional areas thereby

14
14
enabling spatial and temporal crime analysis together with forecasting in an

unsupervised manner. After approximately three years of development, the

system was deployed to the crime analysts during the latter part of 1997. Its

effectiveness is still being evaluated.

DOMESTIC BURGLARY SYSTEMS

One area of volume crime that has attracted the particular attention for

advanced software technologies is that of domestic burglary. Burglary offences

possess special challenges because there are a large number of offences, the same

offender is more than likely to commit a series of such crimes and it is rare to

have a description of the offender.

Devon and Cornwall Constabulary Expert System

During 1985 a pilot system for detecting domestic burglary offences was

developed using a Mimer RDBMS linked to a Prolog interpreter which itself

was used as a query language and as an inference engine [14]. A rule-based

system was developed which matched crimes using a hierarchical arrangement

of modus operandi (MO) after separating domestic from commercial offences.

The MO information was gathered from two viewpoints: -

1. The Officer observations at the scene of the crime

2. From a criminal's perspective

and in six sections: - the disposition of the dwelling before and during the

offence, the method of entry, how the burglary was executed, the method used

to remove stolen items, the items stolen and any special peculiarities.

The MO was codified using a keyword system that represented a path

through the data entry form by ticking nodes from root to leaf. An example

being: -

15
15
 [door , front , ground , secure , lock , forced , bodily_pressure]

Approximately 100 crime complaint reports were processed and data

entered into the system. The rule base comprised two main components 'detect'

and 'extract'. The former used the entire MO set for a crime and constructed

prolog terms that constituted effective search branches with the objective to

identify the most significant MO's and, therefore, to create an MO for

searching against solved crimes. The latter took the created MO and printed a

list of criminals whose MO matched various aspects and on what branches the

matches were made. The criminals were not ranked in any order but listed in

the order in which they were found. The developers established that a

criminal's MO remained reasonably constant across several crimes and a

weighting between 1 and 5 was introduced to reflect the significance of MO

items.

The system was used for three weeks in November 1985 during which

extractions were performed on thirtytwo burglaries. As a result a number of

weaknesses were discovered the foremost being incomplete and/or insufficient

data. The rule base was also enhanced to include the calculation of

consistencies in the MO and the perceived significance of an MO, using these

in the extraction process.

During the live trial no arrests were made for a burglary offence, however,

from the tests made by the developers, indications were that the system had

potential for accurately targeting burglars.

The system was subsequently developed further for the Baltimore County

(Maryland) Police (BCPD) as ReBES (Residential Burglary Expert System)

[15]. Based on 3,800 cases their development produced: - a list of suspects in

16
16
ranked order, messages to aid investigators, MO's that identify suspects

because of some unique behavioural pattern. If, during the matching process,

no match was found or the scale scores too low, the system provided a general

profile of the burglar from the characteristics of the crime.

The Devon and Cornwall system never extended beyond the alpha testing

stage [16] as the concepts could not be proved. Poor data capture was an issue

that was often demonstrated as there were, at that time, insurmountable problems

with the amount and type of data collected. A simplistic crime reporting

system was installed in the Force between mid 1988 to the end of

1989 abandoning the expert system concept. The ReBES system was also

dropped from usage. Reasons given include high turnover in users, new users

disagreeing with the knowledge it contained, the volatility of the knowledge

used by ReBES, the lack of integration with existing computer systems.

Ottawa Police Service

During 1992 and in conjunction with the National Research Council, the

Ottawa Police Service, Canada began development of an expert system to

investigate domestic burglaries [17]. Their overall project objective was: -

To develop a knowledge-based system that will assist the police in the

investigation of residential break & enter incidents that will illustrate the potential for

other applications in support of police investigations.

Eight initial functionality goals were set:

1. provide support for gathering and recording case data,

2. generate suspect characteristics from case parameters,

3. provide the investigator with easy access to:- case parameters, other

incidents in the vicinity, similar cases, patrol officer report narrative, witness

information, sighting reports, investigators' notes

17
17
 4. identify the MO of suspects and cleared cases

5. generate a list of possible suspects

6. identify patterns in the area

7. generate reports as appropriate

8. provide a graphics display capability

The system comprised two main components: data gathering/input and

analysis. The former relied on a paper-based form completed by an Officer,

verified and inputted by an analyst. The latter, a rule based CrystalTM expert

system shell supplemented by C language routines for database access.

Alpha testing took place during the end of 1993 with beta testing

beginning in July 1994. During the former stage an analyst identified five

offences which occurred in the south of the city which had a high probability of

being committed by the same person. An offender was arrested and subsequently

admitted three of these offences.

This system is now being used by a number of the Canadian Police Forces

and is being marketed by a private software company.

Cleveland Constabulary

The current project at Cleveland Constabulary is using a psychological

approach to classify offenders of domestic burglaries into the categories of

'Raiders', 'Invaders', or 'Intruders' by examining the relevant variables of each

crime. A single variable equates to: - an item of property stolen or damaged;

time; date; victim's age; sex; house style etc. There are in excess of 250 such

variables, however only about 90 are relevant to distinguishing the above

categories. Smallest Space Analysis (SSA), a technique first used by David

Cantor [18] in his psychological profiling of criminals, is used to take all of the

18
18
 crime variables and correlate each against the others thereby discovering

similarity ratings.

The results are plotted on a two dimensional graph as illustrated below.

Central Region : this camera

contains the variables antiques
alcohol
that are common to all
three areas such as carrier Raiders
time; video recorder;
front door; broken c/cards
window etc. music
clocks

used > 5 items

watches
facilities cash
insecure tool

ornaments
occupied carried
window
Intruders tool at
forced
scene Invaders
climbed no search

curtains
damage prepare drawn
exit

Figure 3. Smallest Space Analysis Chart

THE IDEAL SYSTEM

Can there ever be an ideal single system? The individual systems within this

paper have each been designed for a variety of uses within its field.

There is a substantial difference between the requirements for recording and

managing crime and its subsequent investigation. Running speculative complex

searches on an on-line transaction processing computer will only slow the system and

detract from its operational effectiveness, therefore, it is good practice for all data to

be transferred into a warehouse for off line processing. Within a warehouse

environment it will be possible to amalgamate data from disparate sources i.e. Crime,

Command and Control, Custody, Courts etc.; enhance the data by creating new fields

based on one or more existing features i.e. banding the time of day that a crime
19
19
occurred etc.; and to provide a consistency of data items across the disparate systems

i.e. road names etc.

Desirable features

It has been shown how investigation systems are being developed to harness to

power of the computer as a tool to aid the solution of both major and volume crimes.

The characteristics of those two categories of crime require different treatments. In the

case of a major crime series it is necessary to search for connections in two or more

crimes when there is a strong prior belief that the same individual(s) committed them.

In the case of volume crime it is very hard for even the most sophisticated computer

user to pick discernible patterns for linking individual crimes. Identifying

characteristics in volume crime where there are no clear suspects is quite difficult due

to specific series of crimes being lost in the morass of all reported crimes.

From the stance of investigating crime, there is a set of features that would be

most desirable in any computer system, the foremost of which is the ability to

interface with disparate, existing systems. It is quite common, at present, to check 4 or

5 different systems to establish the history of and/or intelligence about a single

address or person.

A picture paints a thousand words; visual representation of information

which highlight spatial or temporal connections/patterns is most desirable. Although

there are a number of commercially available packages that excel in this area, it is felt

that the host system should, itself, be capable of such representation.

The ability to use natural language for queries and the facility for any

submitted query having the capability of searching through all tables in the database

without the requirement to establish relationship links. This would bring complex

20
20
searching into the domain of the untrained user. It would also require a complex

indexing system to ensure acceptable retrieval times.

Ability To Visual Automatic AI Techniques Used

Interface Representation Data
Of Data Matching
Major Crime Systems
System
HOLMES No No No No
HOLMES2 No No Yes Yes (inference engine)
Interpol No No No Not Stated
VICAP No No No No
CGT No Yes No ?
Volume Crime Systems
System
CRIS No No No No
CPAS Yes Yes (mapping) No No
AIM Yes Yes (mapping) No No
I2 No Yes No No
Watson Yes Yes No No
SPS No Yes (mapping) ? ?
ReCAP No ? ? Yes (rule based expert
system utilising statistical
analysis)
Burglary Systems
System
Devon & No No Yes Yes (rule based system
Cornwall with inference engine)
Ottawa No No Yes Yes (expert system)
Cleveland No Yes (SSA Yes Yes (fuzzy logic)
charts)
Table 1. System Comparison Table

When data is entered into the system a useful facility would be the automatic

matching of any new information to existing information based on user defined

criteria. Examples being the entry of current burglary offences; the system could

match similar offences based on the MO, geographical area and property stolen and

matching robbery offences by offender description and geographical area. The closest

matches being displayed with a user defined cut off level. The user may require a

90% match on domestic burglaries but be satisfied with a 85% match on robbery

offender descriptions.

The table below offers a comparison of the desirable features associated with

the systems described herein.

21
21
 take in Table 1.

Artificial intelligence (AI) techniques are becoming widely used in a multitude

of commercial products and mainstream computer software applications, examples

being fuzzy logic controllers and neural network pattern recognition. Similar

techniques can be embedded within investigation systems to automatically present

results to the investigator.

DISCUSSION

The computer, as a tool, plays an essential part in the investigation of both

major and volume crime. The majority of current systems are designed to be used in

the investigation of either one of these categories but not both, with the exceptions of

i2 and Watson, and have been written to specifically target a niche market. Each

system performs well when judged against its intended use. However, the systems

mainly rely on users asking specific questions to assist in the investigation process.

Specifically within volume crime, the user may not know which question is relevant

and, therefore, is unable to efficiently link similar crimes in a series.

In both categories of crime it is left to the human investigator to ask the right

questions of the system or to interpret the processed information. This requires

increasing sophistication and specialisation to obtain the best results from the software

and gain confidence in the procedures. However, this is often difficult to achieve in

major crimes due to the relative infrequent occurrence and the voluminous data

therefore the honing of such skills is difficult to accomplish. To achieve best results,

most UK Forces utilise specialist teams of detective Officers who are supported by

trained Civilian Support Staff including trained crime analysts. Such specialism takes

time to develop and may vary in the degree of ability with different aspects of crime.

22
22
 Future systems should now exploit AI techniques that will enable the

automatic examination and presentation of information. However, in the past AI

systems have been viewed with a great deal of scepticism within the Police Force.

This is partly due to a general lack of computer literacy and also due to systems

requiring a substantial amount of hand crafted knowledge with which users may

disagree or the knowledge becoming dated and difficult to update.

23
23
REFERENCES

[1] Sales Literature, Dynamic Reasoning Engine Technology Overview, Cambridge Neurodynamics

Limited, 12th November 1997.

[2] Mansell-Lewis (1997), Shared Intelligence, IBM Police & Justice, pp. 31-32.

[3] Rossmo, K. (1985), Place, Space, and Police Investigations: Hunting Serial Violent Criminals, in

Eck, J. E. and Weisburd, D (Ed.) Crime and Place, Criminal Justice Press, pp. 217-235.

[4] Glaskin, M. (1996) Software tracks down serial killers, Sunday Times, 23rd June 1996.

[5] Atkinson, B. (1996) Cyber cops make house calls, The Globe and Mail, Science and Criminology

17th August 1996.

[6] CRIS Sales Promotion Literature, Crime report Information System, Department of Technology,

Metropolitan Police, London, England, 1997.

[7] Personal interview with Mr. Ian Holden, CRIS Service Manager, Metropolitan Police, 1600 hours

Tuesday 21st October 1997.

[8] Daily Telegraph (1996), Computer Feeling Collars, Daily Telegraph, Tuesday 8th October 1996.

[9] Sales Promotion Literature, Changing the Face of Investigations Around the World i2, Cambridge,

England 1997.

[10] Matthews, R. (1997), Resistance is Futile, Sluth City, New Scientist Supplement, 4th October

1997 pp 20-23.

[11] Sales Promotion Literature, Watson, intelligence analysis, Watson, Cambridge, England 1997.

[12] House, J.C. (1996), Towards a Practical Application of Offender Profiling: The RNCs Criminal

Suspect Prioritization System, Investigative Psychology Conference, Liverpool University 1996.

[13] Charles, J. (1998), AI and Law Enforcement, IEEE Intelligent Systems, January/February 1998,

pp 77-80.

[14] Lucas, R (1986), An Expert System to Detect Burglars using a Logic Language and a Relational

Database, 5th British National Conference on Databases, Canterbury 1986.

[15] Ratledge, E.C. and Jacoby, J.E. (1989), Handbook on Artificial Intelligence and Expert Systems in

Law Enforcement, Greenwood Press, Westport Connecticut.

24
24
 [16] Personal interview with Mr. Paul Lee, Information Systems, Devon and Cornwall Constabulary

1600 hours Friday 3rd October 1997.

[17]Brahan, J.W. Valcour, L. Shevel, R. (1993), The Investigator's Notebook, Unknown source, pp

37-46.

[18] Cantor D. V. (1994), Criminal Shadow Harper Collins, London.

25

View publication stats