Bail Decision Support System PDF

The International Journal Of Engineering And Science (IJES)
|| Volume || 3 || Issue || 8 || Pages || 45-66 || 2014 ||

ISSN (e): 2319 1813 ISSN (p): 2319 1805
Bail Decision Support System

Onyeka Uche Ofili
-------------------------------------------------------------ABSTRACT--------------------------------------------------------
This paper is an improvement on previous work especially the work done by Patricia Hassett It has a more
expended scope and suggests new technological tools and designs for the bail decision support system. It adapts
lessons from effort made in decision support systems for the sentencing domain. And it incorporates new
technological developments such as the neural network into the design of a comprehensive decision support
system for the bail domain. The end result is a robust and novel bail decision support system design, with
detailed systems and user requirements. The decision support system as proposed in this paper should be built
on open architecture, should be easily upgradeable, should have a simple interface, built on web technology
and easy to use.
---------------------------------------------------------------------------------------------------------------------------------------
Date of Submission: 04 August 2014 Date of Publication: 15 August 2014
---------------------------------------------------------------------------------------------------------------------------------------
I. INTRODUCTION
Even though bail decisions seem an obvious choice for decision support systems considering its size (it
is a domain that is not too broad) and the challenges it has faced in the past and in recent times, there is yet to be
robust computer software designed for this purpose. Some of these challenges include; the very limited time
within which a bail decision maker has to reach a decision, wide disparity in bail decisions for similar cases and
lastly lack of documentation of the proceedings and outcome of bail decisions. 1
Notable among the previous efforts in developing support systems in this domain is the work by
Patricia Hassett [1992]2. She developed a prototype bail decision support system focused on addressing the
concern of failure to reappear for summary offences punishable by imprisonment up to six months. Her effort,
which although was a prototype, was not implemented. This may not be unconnected to the often misalignment
between academic researchers in the area of Artificial Intelligence (AI) and law and front-line practitioners. It
could also be that society was not ready for adoption and application of such technology in the bail domain at
the time.
This dissertation will attempt to improve on the work done by Patricia Hassett by expanding the scope
and suggesting new technological tools and designs for the bail decision support system. It will aim to adapt
lessons from effort made in decision support systems for the sentencing domain. And also try to incorporate new
technological developments such as the neural network into the design of a comprehensive decision support
system for the bail domain. The end result is a robust and novel bail decision support system design, with
detailed systems and user requirements.
Section 2.0 of this paper looks at an overview of bail decision support system; meaning of bail, breach
of bail and the challenges the bail system is facing. Section 3.0 analyses some existing sentencing decision
support systems and how they can be used to model a bail decision support system. Section 4.0 looks at the risks
involved in making bail decisions, and subsequently proposes a sound structure and design for bail decision
support system (BDSS). The paper finally concludes in section 5.0.
II. OVERVIEW OF BAIL SYSTEM

In developing a BDSS it is imperative that one understands the bail system including the bail process,
the requisite laws, the consequences of breaching bail and the challenges facing the bail system.
1
P. Hassett, A Prototype Expert System For Making Bail Recommendations, (1992) 7th BILETA Conference,
Information Technology and Legal Education: Towards 2000, 9th & 10th April 1992
2
Ibid
www.theijes.com The IJES Page 45
Bail is the provisional release granted to a suspect while waiting for trial.3 It is an undertaking entered
into by a suspect with a court or police in which the suspect agrees to appear in court at an appointed date and
also comply with every other conditions and terms that may be attached to the bail.4 There are three types of
bail; police bail, police to court bail and court to court bail.5 Police bail is when someone is arrested and later
released due to insufficient evidence for charges to be brought against him. The police can then use the time for
which the accused is on bail to carry out further investigation into the offence before making a final decision.
Before the accused is granted bail he will be interviewed and subsequently be issued a form which will state
when and where to return to answer the bail.6 Police to court bail is when the accused has been charged with an
offence and granted bail to return to the court at a specified date. Court to court bail is when a court grants a
defendant the permission to go and return to the court or another court at a specified date. In applying for bail it
is usually required that the application be made in writing, provide the names and addresses of any sureties, state
the offence, proposed address if granted bail and name of householder.7
Once an alleged criminal is charged he may be released on bail8 except where it is believed that (see Schedule 1
to the Bail Act 1976):
1. There is doubt about his identity and address;
2. It is in his interest and/or the interest of somebody else that he be remanded;
3. There is a strong likelihood that he may abscond and fail to appear in court;
4. There is a strong possibility that while on bail he may obstruct the smooth administration of justice;
and
5. There is a reasonable ground for the court to believe that while on bail he may commit further crime.
Bail may also be denied to a defendant who is charged with murder, attempted murder, manslaughter,
rape or attempted rape. Section 25 of the Criminal Justice and Public Order Act 1994 provides that if the
3
Paralegal Advisory Service (PAS), What is bail? (2004)
<www.penalreform.org/resources/bro-2004-what-is-bail-en.pdf> accessed 16/05/2009
4
Legal Services Commission, What is bail? (South Australia, Law handbook)
<http://www.lawhandbook.sa.gov.au/ch02s03s01.php> accessed 6 June 2009
5
Dorset Police Force, Bail Conditions
<http://www.dorset.police.uk/default.aspx?page=1024> accessed 16 May 2009
6
Patricia M Morgan and Paul F Henderson, Remand decisions and offending on bail: evaluation of the Bail
Process Project (London: Home Office, 1998) Home Office Research Studies
<http://www.homeoffice.gov.uk/rds/pdfs/hors184.pdf> accessed 18 May 2009
7
If a criminal case is brought before the crown Court and the defendant wishes to file for bail while waiting for
the case to reach a conclusion, he will have to apply for bail and fill out form B (Notice of application for bail,
the court of appeal criminal division, Criminal Appeal Act 1968, (Criminal Procedure Rules, r.68.8(2))) as part
of the bail application process. If the application for bail is following grant of conditional police bail see the bail
application procedure as stated in article 84A of the Magistrates' Courts Act 1980 (Magistrates' Courts
(Amendment) Rules 1995).
8
Section 4 of the Bail Act 1976 provides that bail shall be granted to:
4(2) (a) To a person who appears or is brought before a magistrates court or the Crown Court in the course of
or in connection with proceedings for the offence, or
4(2)(b) To a person who applies to a court for bail or for a variation of the conditions of bail]in connection with
the proceedings.
4 (3) To a person who have been convicted of an offence, appears or is brought before a magistrates court to be
dealt with under [Part II of Schedule 3 to the Powers of Criminal Courts (Sentencing) Act 2000 (breach of
certain community orders)].
4(4) To a person who has been convicted of an offence and whose case is adjourned by the court for the purpose
of enabling inquiries or a report to be made to assist the court in dealing with him for the offence.

defendant has previously been convicted of any of the above mentioned crimes in the past he will not be granted
bail except for very exceptional grounds. This has been thought to be in conflict with the provisions of Article 5
of the European Convention on Human Rights. However, it was clarified that these two provisions are not in
conflict in R(O) v Harrow Crown Court (2003) The Times, 29 May by Kennedy LJ.9
Bail may be granted with conditions or without conditions attached. The police and court can impose
conditions to bail granted for several reasons. It could be to ensure that the accused returns to court at an
appointed date, to ensure that the accused does not go out and disrupt or interfere with witnesses whiles on bail
or to prevent the accused from committing further crime. Some of the conditions imposed could be to restrict the
accused movement by asking him not to move out of the town where the crime was committed or asking him
not to move within certain distance to a witness residence or office. Curfew may be placed on him requesting
that he does not stay outside his house beyond a given time and does not leave his house before a set time. On
the other hand the court can grant an unconditional bail to a defendant where and when it is convinced that the
defendant will return to court at an appointed time. And that while on bail he will not try to disrupt the
administration of justice on that particular case or indeed any other case for that matter and that he will not
commit further crime.10
2.1 Breach of Bail

It is an offence for a person to fail to show up in court on the agreed date or violates any of earlier
stated bail conditions where applicable. For instance, it is an offence for the person when asked not to travel
outside a given geographical location to do so without the knowledge and consent of the court. If the person is
found guilty of not adhering to the terms and conditions of bail it may lead to his immediate arrest and the bail
withdrawn. The person may end up being remanded in custody and may not be granted bail in future or stricter
conditions may be imposed on him. It is generally considered a criminal offence to breach bail. Section 6 of the
Bail Act 1976 subsections 1and 2 provide that: (1) if a person who has been released on bail in criminal
proceedings fails without reasonable cause to surrender to custody he shall be guilty of an offence; (2) if a
person who has been released on bail in criminal proceedings, and having reasonable cause therefore, has failed
to surrender to custody, fails to surrender to custody at the appointed place as soon after the appointed time as is
reasonably practicable he shall be guilty of an offence.
In a situation where the accused believes he has good enough reason that prevented him from
surrendering to custody, s6 (3) of the Act provides that it shall be for the accused to prove with substantial
reason that he has justifiable reason for not surrendering to custody. Where a person is found wanting with
respect to subsections 1 and 2 of Section 6 the offence is punishable under subsection 5 of the same section,
either on summary conviction or as if it were criminal contempt of court.
2.2 Challenges of the Present bail System

Having a functional and effective bail system is very important in every society. The bail system
should not be too lax neither should it be too rigid and insensitive to allow innocent citizens suffer under
detention. The challenge therefore, is to have a bail system that carefully marries these two concerns together; a
system that is able to reduce risk to the public while protecting the rights of the citizen, including the right to
freedom unless convicted.
The then shadow Justice Secretary, Nick Herbert, in 2008 suggested that the bail system was too
weak.11 The conservatives generally argue that the public interest be given more attention in bail decisions. This
view is in line with the call by Scotland Police chiefs in May 2005 on the then Justice Minister, Cathy Jamieson,
to get tougher on criminals who commit crimes while on bail. At the time it was recorded that more than 21,000
offences have been committed by people on bail in the Strathclyde Police area alone within a period of one
9
Robert Jago, Civil and Criminal Procedure
<http://www.londonexternal.ac.uk/current_students/programme_resources/laws/subject_guides/civ_crim/crimin
al_procedure_ch11.pdf> accessed 8 June 2009
10
Criminal Justice System, Upholding the Rights of the Defendants
<http://www.cjsonline.gov.uk/defendant/bail/> accessed 18 April 2009
11
Michael White and Andrew Sparrow, Conservatives unveil plan to reform bail system Guardian (London 11
August 2008)
<http://www.guardian.co.uk/politics/2008/aug/11/justice.conservatives> accessed 17 May 2009

year.12 Further evidence in support of this view can be seen in the report that 60 out of the over 450 (that is
about 13 percent) murder suspects charged in 2008 were bailed after being charged. And also four out of every
five violent crime suspects gets bail13.
Nick Herbert was of the view that the fee of 60 for failing to answer bail was too small and
insufficient for not showing in court at the due date. He further thinks that defendants are easily granted bail,
bail is breached frequently and the enforcement of bail rules is wanting. Having a lax bail system is against the
public interest as the criminals who are easily let go free may go back to commit more crime while on bail. This
may affect the confidence which citizens have on the general judicial system. Gordon Brown instigated that the
bail policy should be reviewed following the killing of his mother-in-law and himself by a former police
inspector, Garry Weddel, while out on bail.14 If this incident is anything to go by one would recommend a
stricter bail policy and possibly having a blanket bail system. However, having a blanket bail policy, will result
to breach of human right and go against the principles of human right laws15. Article 5(1)( c) of the European
Convention on Human Rights and Fundamental Freedoms provides that everyone has the right to liberty and
security and no one shall be deprived of this right except in the following instances: the lawful arrest or
detention of a person effected for the purpose of bringing him before the competent legal authority on
reasonable suspicion of having committed an offence or when it is reasonably considered necessary to prevent
his committing an offence or fleeing after having done so. Furthermore, Article 6 of the Convention provides
that everyone has a right to fair trial. Bail decisions should therefore, be treated on a case by case basis.
The judicial system as pointed out above should not adopt a blanket bail policy. Remanding people
unnecessarily will also put more pressure on the prisons. Overcrowding of prisons had already become a major
headache to the UK government. [16][17][18] Unfair detention of defendants and over crowding of prisons has been
reported to be one of the causes of prison suicide in UK. 19 The bail system is partly blamed for this problem.
12
The Journal Online, Bail system in crisis, say police chiefs: Conference call to Executive for crackdown on
bail offenders The Journal Online (London 20 May 05)
<http://www.journalonline.co.uk/News/1001851.aspx> accessed 4 July 2009
13
Michael White and Andrew Sparrow, Conservatives unveil plan to reform bail system Guardian (London 11
August 2008)
14
Ibid
15
The Law Commission, Criminal Law: Bail and The Human Rights Act 1998: A Summary (1999)
Consultation Paper NO 157
<http://www.lawcom.gov.uk/docs/cp157sum.pdf> accessed 17 May 2009
16
BBC News, Prison overcrowding 'at crisis point' BBC News (London Wednesday, 28 August, 2002)
<http://news.bbc.co.uk/2/hi/uk_news/2222022.stm> accessed 20 May 2009
17
Sophie Goodchild, Public at risk from prison overcrowding The Independent (London Sunday, 21 January
200)
<http://www.independent.co.uk/news/uk/crime/public-at-risk-from-prison-overcrowding-433075.html>
accessed 20 May 2009
18
Alan Travis, Overcrowding blamed for rise in prison deaths The Guardian (London Saturday 22 September
2007)
<http://www.guardian.co.uk/uk/2007/sep/22/ukcrime.prisonsandprobation> accessed 20 May 2009
19
Nigel Morris, Rise in prison suicides blamed on overcrowding The Independent (London Friday, 17 June
2005)
<http://www.independent.co.uk/news/uk/crime/rise-in-prison-suicides-blamed-on-overcrowding-494408.html>

The unnecessary remand of defendants is seen as one of the major causes of the rise in the prison population.20
Juliet Lyon, Director of the Prison Reform Trust, said: "This catalogue of deaths must act as a terrible warning
to the courts to avoid custody for those who are vulnerable or mentally ill, and a stark wake-up call to
Government to act now to improve court diversion, bail provision and treatment options and, at long last, to end
prison overcrowding."21
Another challenge facing the bail systems is the fact that magistrates have very limited time within
which to make a bail decision. This leads to inconsistency in bail decisions (a major human right concern, see
above), lack of proper documentation and no time to adequately reference past similar instances. The issue of
inconsistency in bail decisions is not only peculiar to this domain but also to the sentencing domain, a domain
which has some similarities to the bail domain. In fact U.J Schild (1998) argues that in the domain of criminal
sentencing that it is possible for a judge to arrive at different conclusions even under identical conditions. It all
basically depends on what the judge has at the back of his mind to achieve. The judge may arrive at a different
conclusion if rehabilitation is what he has in mind and may arrive at a different conclusion in terms of
sentencing if deterrence is what he has in mind.22 Furthermore, K. White (2004) argues that similar crimes may
end up with different sentences and one factor that has to be taken into consideration is societal expectations.23
This situation is quite similar to what obtains in the bail domain. A research carried out in 2004 in New York
using courtroom observations complemented with the states Criminal Justice Agency databases revealed that
some factors influence the bail decision reached by judges24, see section 4.1.2 below. In reaching his decision
the judge may not only consider the likelihood of the defendant fleeing but also the possibility of pre-trial crime
or he may choose to deny the defendant bail as a mark of pre-trial punishment.25
The foregoing arguments are not out of context but there should be more consistency in decisions
reached by judges including bail decisions. This can be achieved by the use of decision support systems
specifically suited for the bail domain. This system will also provide judges with accurate information and help
them make faster and more informed decisions. This decision support system should in no way impair the
freedom or independence of judges. 26 And it does not imply that judges will no longer be subjective on matters
pertaining to bail but they will carry out bail decision based on facts and accurate information made readily
available to them through the bail support system. 27 And even though the sentencer may have the discretion and
20
Prison Reform Trust, Five ways to stem prison overcrowding, October 2005
<http://www.prisonreformtrust.org.uk/subsection.asp?id=349> accessed 20 May 2009
21
Nigel Morris, Rise in prison suicides blamed on overcrowding The Independent (London Friday, 17 June
2005)
<http://www.independent.co.uk/news/uk/crime/rise-in-prison-suicides-blamed-on-overcrowding-494408.html>
22
U. J Schild, Criminal sentencing and intelligent decision support, (1998) 6 Artificial Intelligence and Law,
volume 6: 151
23
K. White, From Knowing to Legal Knowledge: Using Early Twentieth Century Canadian Murder Trials to
Problematize Knowledge Management Technology (2004) CAUT Law Forum - Winnipeg, 2004
24
Mary T. Phillips, Release and bail conditions in New York (2004) New York City Criminal Justice Agency,
Research brief No. 6, August 2004
<http://www.cjareports.org/reports/brief6.pdf> accessed 7 June 2009
25
John S. Goldkamp and Michael R. Gottfredson, Bail decision making and pre-trial detention Surfacing
judicial policy (2005) Springer Netherlands, Law and Human Behaviour
26
U.J. Schild, "Criminal sentencing and intelligent decision support", (1998) Artificial Intelligence and Law,
volume 6: 151

the last say in deciding whether a suspect may be granted bail or not it is an offence for him to do so with the
wrong motives, maliciously, unlawfully or without substantial reason or probable cause, see Linford v Fitzroy
(1849) 13 QB 240 at 247; R v Badger (1843) 4 QB 468 at 472 and Osborne V Gough 3 B & P 551.
2.2.1 Problems with the Bail Act

Schedule 1, part 1 Article 2 of the Bail Act, 1976 provides that a defendant should not be granted bail if
the court is satisfied that there are substantial grounds for believing that the defendant if released on bail would
fail to surrender to custody, commit an offence while on bail, and interfere with witness or otherwise obstruct
justice be it in relation to himself or someone else. The Act did not clarify on what is meant by substantial
grounds for believing. It did not in itself provide in explicit terms what constitutes relevant grounds for
believing. The law commission have argued that requirements as to what constitutes substantial grounds for
believing should be added to the Act.28 There is no body of judicial rules which one can point at that supports
or provides reasonable explanations for determining grounds for believing. It is unlike other aspects of law
where history of judicial rules provides substantial guidance. However, efforts are been made to address this sort
of gaps in legal documents, ranging from the use of mathematical models29 to Larry Laudans reform proposal
which suggests that focus should be on objective evidence in place of jurors subjective viewpoint.30Some of
these proposed solutions could be adopted for the bail domain.
The UK legal system relies on case-based reasoning as a guide to determine present and future cases31.
This is referred to as the doctrine of stare decisis. That is the outcome, rulings, in previous cases of similar
nature are used as guide to decide present cases, invariably like cases should be treated alike. Even though bail
decisions are not usually discussed in the context of stare decisis, the same moral perception applies: courts
deviating without reason in their bail decisions from established practice violate one of the most fundamental
tenets of justice.
Case-based reasoning for now is not as effective in bail system as it is in other legal domains. The
reason for this include the fact that in bail decisions, there are no mechanism for properly documenting the
proceedings and decisions reached and the reasons for which such decisions were taken. The time frame from
start to finish for a bail case is usually very short. And the courts have not really had reasons enough to put in
place mechanism for documenting the proceedings perhaps due to time constraint and the volume of bail cases
that need to be addressed within the very limited time.
Part of the principles of the stare decisis doctrine is the respect and regard given to past judgements
made by higher courts (appellate courts) over lower courts. A lower court is bound by the decisions of a higher
court. In bail systems it is not very often that you see bail decisions appealed to higher courts. The reasons are
still within the major challenges of the bail system, the time for hearing a case is too short and there is no proper
documentation. For a decision to be appealed you need to show convincingly that the decision reached by the
lower court lacked merit.
In the absence of rules on a particular subject by a higher court, a lower court can decide to rely on
previous rules by another court of equal level. Such reliance by a court on another court of equal status is
27
Samantha Besson, Four Arguments Against Compromising Justice Internally, (2003) Oxford Journal of Legal
Studies, 1 June 2003
28
The Law Commission, Bail and the Human Right Act 1998: Executive Summary LawCom No 269,
<http://www.lawcom.gov.uk/docs/lc269sum.pdf> accessed 5 June 2009
29
Thomas F. Gordon, Henry Prakken and Douglas Walton The Carneades Model of Argument and Burden of
Proof (2007)
<http://www.dougwalton.ca/papers%20in%20pdf/07GordonPrakkenWalton.pdf> accessed 4 August 2009
30Larry Laudan, Truth, Error, and Criminal Law: An Essay in Legal Epistemology, (Cambridge Studies in
Philosophy and Law, CUP, Cambridge 2005)
31
Sharon Hanson, Legal method and reasoning, (2nd Edition Cavendish Pub Ltd, London 2003)

usually motivated by the degree of persuasiveness of the rulings in the case. But the same challenge will be
encountered within the bail domain. Even though courts at the same level may have made bail decisions on
similar matters there may be no documentation on the outcome of those cases.
This lack of substantial record of past cases is a major challenge to the construction of a robust expert
system for the bail domain. Although, this challenge is not impossible to overcome in building an expert system
for the bail domain but it must be taken into consideration by all means in designing such system.
III. SENTENCING DECISION SUPPORT SYSTEMS

Not too much work has been done in the area of developing expert systems or decision support systems
in the bail domain; in fact the effort in this area is very sparse. This is however, not the case for the sentencing
domain even though it shares a lot in common with the bail domain. This section will start by looking at the
general challenges facing the sentencing domain. It will then go further to look at the progress made and
challenges encountered in designing and deploying decision support system in the sentencing domain. It will
also consider how the lessons learnt from the sentencing domain can be adopted and applied in the design of a
decision support system for the bail domain. Since we will argue that it is possible to transfer most if not all of
the ideas from sentencing support to bail decision support, we will now discuss sentencing support systems in
some length starting with the challenges facing the domain.
One of the challenges facing the sentencing domain is the complaint that there are a lot of disparities in
the sentencing decisions made by judges. And this is attributed to the plenty of room given to judges to use their
initiatives and the challenge posed by huge statutory provisions which the judges have to come to grips with.
Another problem is variation in content and scope and variation in quality of the information presented before
the Court and for which decision has to be based on.32 Other reasons for lack of uniformity in sentencing are due
to different sentencing aims and different judges assigning varying weight to each sentencing aim. These
challenges as is in the sentencing domain are very similar to the challenges in the bail domain see above.
In solving the earlier pointed challenges in the sentencing decision system various jurisdictions have
adopted strategies and methods which they consider suitable for their particular society. Some states in the US,
such as Minnesota, for example have adopted sentencing guidelines.33 The US Sentencing Reform Act of 1984
forms the foundation for the Federal Sentencing Guideline34 as seen today. The guidelines provide sentencing
ranges for various crimes. However, in U.S. v. Booker, 543 U.S. 220 (2005), the Court held that sentencing
guidelines as constituted violated the Sixth Amendment right to trial by jury. It also held in the second opinion
that the sentencing guidelines should be treated and used as discretionary or strictly for advisory purposes. This
decision goes to restore to judges the power to use their discretion in passing a sentence on a given case and
saving the system from been purely mechanistic.35
Other jurisdictions are seeking the use of computer systems to provide decision support for judges or a
combination of decision support systems and sentencing guidelines. Sentencing decision support systems as
implemented in some of these jurisdictions (see below) are not introduced with the purpose of replacing judges
but rather as a mere support to judges to help them more efficiently perform their duties. A judge is still left with
the right to heed to the recommendations of the computer systems in part or in full or even discard it entirely.
There is room for judges to use their own discretion (been able to choose from a list of correct answers only this
time the correct choices are backed with substantial analytic reasoning and evidence provided by the computer).
32
David Bainbridge, CASE': Computer Assisted Sentencing in Magistrates' Courts (2005) 5th BILETA
Conference
33
Andrew Von Hirsch, Sentencing guidelines and penal aims in Minnesota (1994) Criminal Justice Ethics vol.
13, 1994
34
United States Sentencing Commission, Federal Sentencing Guidelines Manuals (2008)
<http://www.ussc.gov/2008guid/GL2008.pdf > accessed 3 June 2009
35
Lisa M. Seghetti and Alison M. Smith, CRS Report for Congress, Federal Sentencing Guidelines:
Background, Legal Analysis, and Policy Options Updated June 30, 2007
<http://www.fas.org/sgp/crs/misc/RL32766.pdf> accessed 3 June 2009

However, such systems will to a very good extent reduce the disparity in the sentencing decisions as it will
provide in clearer terms to a judge how a decision is arrived at based on statues and past cases of similar nature.
Sentencing decision support system or sentencing information system (SIS) as it may be referred to in
some parts of this paper have been around for over two decades. Its level of development and adoption varies
from one jurisdiction to another. Some of the countries that have embraced it include Canada, Scotland, Israel,
England and Wales, Netherlands, and Australia. Amongst these jurisdictions Canada was the first reported to
have experimented with SIS. She however, recorded very limited success (Doob and Part, 1987).36 The same is
not particularly the story in other jurisdictions, as the few jurisdictions that have tried out SIS have recorded
varying level of success in terms of the design of the system and its subsequent application.
ENGLAND AND WALES

Computer Assisted Sentencing (CASE) was built for sentencing in Magistrates Court in England and
Wales due to the diversity of the cases that are brought before the court. Magistrates are lay persons with little or
no legal background and it is the first court majority of offenders are brought to.37 The court in passing a
judgement on a particular case may be faced with a variety of correct answers. The court is bound to choose
from these right answers but not arbitrarily, there must be a strong basis for reaching a particular decision or
picking a given choice. This is a huge challenge as the court will have to take into consideration the uniqueness
of each case and apply the most appropriate principles.38 This necessitated the development of CASE to assist
magistrates and help make their work easier.
The system was built using Microsoft BASIC after unsuccessful attempts to use existing shells, such as
Micro Expert and CRYSTAL II. Shells can be quick and effective for developing sentencing system and indeed
any software system. They however, have the disadvantage of constraining the development of the system to the
shells framework or existing structure. The proposed bail decision support system design which will be
discussed later will not be built using any existing shells due mainly to the disadvantage pointed out above.
ISRAEL
HaCohen-Kerner & Schild (2001) built an SIS called the Judges Apprentice. It is a case-based system
and provides support to judges in the sentencing of rape and robbery crimes. The system is like a tree with each
leave on the tree representing an index. There are 371 leaves or legal concept, each relevant to specific criminal
sentencing. These indexes are used for establishing index similarity between the case at hand and previous cases
within the sentencing tree. The system retrieves similar cases and also helps in the selection of the most suitable
case among the bunch. After which a case-based quantitative assessment is done and used as a basis for arriving
at a verdict for the case at hand.39 In building a BDSS a good starting point will be to begin by limiting the use
of the support system to some selected crimes as a pilot. And then monitor the progression and performance of
adopting such a scheme and then gradually move it to other forms of crime having tested and debugged the
system. The support system can even be modelled to use an index tree similar to the design of the Judges
Apprentice.
NETHERLAND
This system is designed for the northern part of the Dutch Judicial System, it is called NOSTRA. It is
limited in scope to offences with not too complex sentencing decisions. Its architecture is such that more
features can be added to it with time- that means it has an open architecture. It provides judges with the ability
to compare present cases with previous cases, see the decision in the previous cases and see the argument
preceding the decision. The actual programming language used in developing NOSTRA was not clearly stated
36
A.N Doob and N.W Park, 'Computerised Sentencing Information for Judges' (1987) Criminal Law Quarterly
30: 5472
37
David Bainbridge, 'CASE': Computer Assisted Sentencing in Magistrates' Courts (2005) 5th BILETA
Conference
38
These intricacies and the careful and rightful application of the law is seen in the judgement of Dunn L.J. in
De Havilland(1983) 5 Cr. App. R. (5) 109 and the judgment of Lord Lane C.J. in Barrick (1985) 7 Cr. App. R.
(5) 142
39
Y. Hacohen-Kerner and U.J. Schild, Case-based Sentencing Using a Tree of Legal Concepts, (2001)
Information & Communications Technology Law, Volume 10, Issue 1 March 2001 , pages 125 - 135

by Jenne et al (1998). Although it was stated that it was built on open platform and it is a case-based decision
support system.40 One key lesson from this system which is applicable to the proposed BDSS is its open
architecture. The proposed BDSS will be built on an open architecture thus having the provision for future
features to be added to the system as the need arise and able to readily interconnect to other systems.
NEW SOUTH WALES, AUSTRALIA

The New South Wales SIS was developed primarily not to curtail discretionary judgement by judges
but rather to provide more and accurate information upon which a judge can make a more informed
discretionary judgment. And of course like other SIS it aims to also promote consistency, rationality and at the
same time assisting judges to adhere to the tenets of the law.41 This SIS is built using web (HTML) technology,
meaning it can be accessed using regular web browsers such as Netscape and internet explorer. The application
subsystem is built on Windows NT, MS IIS Server, Netscape Communication Server, ISYS.web Search engine,
Topic Search Engine and MS Office Professional. The SIS can be extended to include more features because its
architecture is open allowing for easy expansion. It can seamlessly be interfaced with other systems that can
generate ASCII or other standard word processing output. And importantly it can be connected to other existing
databases through Open database Connectivity (ODBC).42 This is important because it may become necessary
for the SIS to be linked to other databases that may hold vital information on citizens from which a judge can
obtain certain vital information on offenders. The proposed bail DSS will be built using web technology such
that it can be accessed over the web from a remote location using regular web browsers such as internet
explorer, Mozilla Firefox, Netscape and chrome. And its database structure and design will be such that it can
easily be connected to other databases.
SCOTLAND
The Scottish SIS basically took its lead from that of New South Wales, Australia. It all began after
Lord Justice Clerk got an inspiration from the demonstration of the New South Wales System in a conference of
the Commonwealth of Learning held in Canada.43 Work started in earnest on the Scottish SIS in 1993 in the
University of Strathclyde. But the system did not go into full use in the High Court until February 2002.44
The SIS comprises of two main subsystems. The first subsystem is the interface that allows for data to
be entered into the system thereby updating the databases. With this judges and clerks can enter new cases into
the system. It also allows judges to enter information detailing the reasons behind the decision they took on a
particular case. The second subsystem is for data retrieval. The system allows judges to readily retrieve data on
a previous case entered into the database. The system more or less has the same set of features as is the case
with the New South Wales SIS. The proposed bail DSS will have high security and allow different levels of
access, while some may have read only access some may go as far as having both read and write access. People
with write access such as clerks and judges will be able to update the database from time to time. However,
when the database is updated before the new addition is finally accepted it will go through a set level of
approval. This is to create checks and balances so that the systems is not abused and wrong data entered or
deleted without following due process.
40
Jenne Van Der Vinne, Ing. W Van Zwol and M Karnekamp A Sentencing Information System Named
'NOSTRA' (1998) IJL&IT 1998 6 (230)
41
Austin Lovegrove, Statistical Information--Systems as a Means to Consistency and Rationality in
Sentencing (1999) IJL&IT 1999 7 (31)
42
Ibid
43
Cyrus Tata, John N. Wilson and Neil Hutton, Representations of Knowledge and Discretionary Decision-
Making by Decision-Support Systems: the Case of Judicial Sentencing (1996) JILT 1996 (2)
<http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/1996_2/tata/#a3.2.4.1> accessed 2 June 2009
44
The Sentencing Commission for Scotland, The Scope to Improve Consistency in Sentencing Report 2006
<http://www.scottishsentencingcommission.gov.uk/docs/consistency/Consistency%20Report%20-
%20Final.pdf> accessed 30 May 2009

The proposed BDSS will be built taking into consideration all the above mentioned lessons from
existing sentencing decision support systems and indeed the work by Patricia Hasset on BDSS.
IV. BAIL DECISION SUPPORT SYSTEM

Hassetts work, although a prototype, is quite insufficient and limited in scope to solve the challenges
of the bail system. A system that checks if an offender will answer bail when released is not robust enough to
solve the present shortcomings of the bail system or provide substantial help to judges in making bail decisions.
A defendant should not be denied bail only on the basis that he may not answer bail nor should he be granted
bail only on the premise that there is substantial evidence that he will not flee. It is not that this is not a good
ground to deny a suspect bail but it should not be the only ground for which a defendant is denied bail.
Schedule 1 to the Bail Act 1976 provides other conditions under which an accused may be denied bail. Some of
these include if there is a strong proof that the accused may commit an offence while on bail, interfere with
witnesses or otherwise obstruct the course of justice, whether in relation to himself or any other person and if
the court is satisfied that the defendant should be kept in custody for his own protection or, if he is a child or
young person, for his own welfare.
If there is strong suspicion that an accused will fail to surrender to custody if granted bail, he may be
granted bail with strong conditions. His travel passport may be ceased. He may be asked not to move outside a
particular locality. His movement may be monitored by responsible law enforcement body. So if the defendant
can be prevented from not reporting to the court why deny him bail simply because there is some doubt that he
may not return for trail.
However, if there is some other supporting evidence against the accused such as one or more of the
exceptions under Schedule 1 to the Bail Act 1976 the accused may be refused bail. Take a scenario where an
accused is granted bail on the fact that he will definitely return to the court on the appointed date but failing to
recognise that the accused may go ahead to commit further crime when on bail as was the case with Garry
Weddel see above. Likewise the fact that a defendant has past criminal record does not mean that if released he
will go back to commit further crime. The weight of the crime or the severity of the likely sentence against the
defendant if found guilty is not sufficient to conclude that the defendant will abscond if granted bail. In other
words other factors have to be taken into consideration when making a bail decision.
To ensure that judges render the right bail sentence taking into consideration the provisions of the bail
Act, there is need to seek the help of intelligent computer systems, a system that will focus on the court not
denying or granting a defendant bail unjustly. This system is a decision support system tailored to the bail
domain. A BDSS is, therefore, computer software designed to assist judges and police-officers make the right
bail decisions. Decision support systems are expert systems if they do not stop at presenting raw data to the user.
If the system goes the extra mile of analysing the data and providing the user with suggestions as to how it
arrived at a conclusion then it is an expert system as it possesses some intelligence. Take an example of a judge
who was presented with the case of a suspected criminal who apparently is seeking bail until the next hearing of
his case in court. Obviously the judge will want to look up the suspects past criminal record, consider the
circumstances surrounding this very crime, check if the suspect has been granted bail for any previous offense
and whether he kept to the bail conditions. He may also want to refer to the outcome of previous cases of similar
nature and circumstance if available. When the judge keys in the personal details of the suspect the system
should be able to immediately query its database and provide the judge with the suspects past criminal records.
It should also provide the judge with other details such as the marital status, employment status and age of the
suspect. The system will then use these two sets of details to provide recommendation as to whether the suspect
should be granted or denied bail and the possible bail terms and conditions. The system should also be able to
provide reason for its recommendations, thus doing the hard work for the judge. The judge does not have to take
the recommendations of the BDSS verbatim but it can be a very solid guide to the bail decision to be made by
the judge. This system should be able to seamlessly connect to other existing relevant databases and extract
information based on the judges query that are relevant to specific cases.
4.1 Risk Assessment: Reasons for Developing a BDSS

The decision maker has to ascertain if there is substantial grounds to believe that the accused will not
reoffend, flee or obstruct justice if released. If substantial grounds does not exist for the decision marker to
believe that the accused will reoffend, flee or obstruct justice if released, the statues provide the accused is to be
released and asked to return to court at a particular date for hearing of the case. If substantial grounds does,
however, exist for the decision maker to believe that upon release the accused may flee, obstruct the

administration of justice or re-offend, the statues provide that the accused should be detained or released with
conditions. The bail decision maker has to carefully analyse the risk of releasing or detaining an accused.
Experienced magistrates may be able to use their wealth of experiential knowledge to determine if an
accused will flee or reoffend if granted bail based on certain considerations but the chances of the magistrate
getting it wrong is very high. However, this can be more efficiently achieved using data mining tool that
correlates records in the criminal justice system with personal sociological features. In addition statistical
analysis can be carried out on the behaviour of previous cases of bail granted or denied. The analysis will help
determine how many persons granted bail fled or re-offended within a given time period. It will show what kind
of crime has the highest and least number of re-offenders, accused that fled and those that tampered or attempted
to obstruct the administration of justice. It will show how many of these people are female or male, their ages
and other vital details. This sort of statistical tool will be an integral part of a robust BDSS such as the one
proposed in this paper.
Yet another form of risk worth giving full attention is the risk of wrongfully denying bail or granting
bail. As mentioned earlier it is improper and very unfair to an accused if he is wrongfully denied bail. This may
cause the accused to suffer various devastations such as relationship with family, loss of employment, health
issues and physiological trauma. And on the other hand, as also earlier pointed out granting bail wrongfully can
affect the confidence the public has on the judicial system. The person released on bail can go back and commit
further crime (see the case of Garry Weddel above) or flee. It is therefore, very important that in-depth risk
analysis be carried out before granting bail or denying an accused bail. This sort of analysis cannot be carried
out immediately by the decision maker within the very limited time frame which he has to make a bail decision
without needing the help of a well designed and tailored software tool. The BDSS proposed in this paper will be
able to carry out this risk analysis and present the decision maker with results and reasons behind each result.
This still allows the decision maker room to use his discretion to choose the most appropriate answer but this
time he sees the reasoning behind each result, he does not have to choose a result and corresponding reason but
may modify the result and/or the reasoning behind the result to more suitably address the case at hand.
Finally, it may be wrong for similar crimes committed under very similar circumstances to attract bail decisions
that are very wide apart. The bail process as presently structured does not have what it takes to fix this
challenge. It will take a system that has a database of previous crimes and their corresponding bail decision to
solve this problem.
4.1.1 Risk of Reoffending

A study titled Re-offending on Bail in Avon and Somerset (Bristol: Avon and Somerset
Constabulary, 1991) revealed some interesting results on factors that could influence the likelihood of an
accused reoffending while on bail. It showed that 28 percent of accused granted bail reoffends from a court
summary, CID survey showed the proportion to be 27% while the questionnaire result from Custody Officer
showed the portion to be 12 percent. According to the study the major factors associated with reoffending
include;
Age of defendant: younger defendants (those aged between 17 and 20 years) are twice more likely to
reoffend than those aged 26 years and above,
Type of offence: suspects on bail for vehicle-related crime and burglary showed high likelihood to reoffend,
and
Number of previous convictions has also shown to be a major contributor to the likelihood of reoffending.
A more recent study than the one above, Offending on Bail and Police use of Conditional Bail
(Brown, D., Home Office Research and Statistics Directorate, Research findings No. 72, London: Home Office,
1998, p.1) gave revelations that were not too far from the previous study. It showed that suspects granted bail
for vehicle-related crime have the highest tendency of re-offending (44 percent) followed closely by suspects on
bail for crime of theft by shoplifting (40 percent). It also revealed that younger people (juvenile) are twice as
likely to reoffend compared to their adult counterparts. Another study Remand Decisions and Offending on
bail: Evaluation of the bail Process Project (Morgan, P., and Henderson, P., Home Office, 1998, p.45) revealed
that the following factors are also responsible for re-offending: persons with no fixed home address (42 percent),
suspects who had to wait for more than six months before trial or sentence (32 percent), suspects charged with
car theft (32 percent), suspects who has previously breached bail (27 percent), those who have previously been
jailed (28 percent), suspects below 18 years of age (29 percent) and unemployed suspects (21 percent). This
study also went further to show that suspects with shorter period to trial disposition (that is waited for less than
say one month) were less likely to reoffend (4 percent). Also only 7 percent of employed suspects will reoffend

and only 6 percent, 7 percent and 8 percent respectively for sex offences, assault and fraud. From the foregoing
the risk of reoffending can be summarised as shown below:
High: Suspect is below 18 years of age; Suspect has no fixed address; Offence is vehicle-related or burglary;
Suspect is not employed; Suspect has previously breached bail; Suspect has gone to jail before for a crime;
Suspect has to wait for more than six months for case to be heard.
Low: Suspect is adult, above 18 years of age; Suspect has a fixed home address; Suspect has to wait for a short
period before trial; The offence is sex related, assault or fraud.
4.1.2 Risk of non-appearance

Ensuring that a suspect will attend his trial as at when due is very important in the granting of bail. This
was affirmed by Lord Russell in R v Rose (1898) 78 LT 119. According to a research carried out by New York
City Criminal Agency some factors should be taken into consideration when assessing risk of flight. These
factors include; family ties, probable sentence (strength of offence), past criminal record, employment status,
how long suspect has lived in the community, character and mental stability of suspect and whether or not
suspect has failed to surrender to bail in the past.
Patricia Hassett (1992)45 pointed out that risk of flight is influenced by whether the penalty upon
conviction is custodial or not and also by the level of family and community ties. A man who has a wife, with
children who are in school, has a mortgage of which he has paid substantial equity and has a good and stable job
is less likely to abscond. He will not want to lose all the above mentioned simply because he wants to avoid a
short period of incarceration.
A study showed that 7 percent of defendants granted police bail failed to attend first court appearance
as at when due and 9 percent of those granted court bail failed to attend at least one court hearing. (Brown, D.,
Offending on Bail and Police use of Conditional Bail(1998) Home Office Research and Statistics Directorate,
Research findings No. 72, London: Home Office, 1998, p.1). Neil Corre and David Wolchover (2004)46 pointed
out that factors which have likely influence on whether a subject will abscond include; nature and seriousness of
offence, the character of the defendant (has he been convicted of previous offences if so how many times, see R
v Vallet [1951] 1 All ER 231), the kind of people he associates with, community ties, previous bail history,
character, the defendants mental stability, family ties, probable sentence (strength of offence), past criminal
record and employment status. From the foregoing the risk of non-appearance can be summarised as shown
below:
High: No strong family ties, perhaps say not married or have any kids; Does not have a fixed address or
any form of ties to the community; Has previously breached bail; Has previously been convicted on a related
offence; Penalty if convicted is custodial; Has no job; Is linked with a criminal group; Is not mentally stable.
Low: Strong family ties; Employed; Strong community ties; Never breached bail in the past; Penalty if
convicted is not custodial; No past criminal record; Mentally stable and of good character
4.1.3 Risk of interfering with the administration of justice

The risk of obstruction to the proper administration of justice is one that must be well assessed in
determining whether a defendant should be granted bail. Witnesses must not be intimidated or coerced in any
way and evidence must not be tampered with by the defendant. Factors that should be considered in measuring
the risk of interference by a defendant include;
The relationship between the defendant and the witness. If the defendant is very close to the witness,
say they live in the same house, chances are that the defendant may interfere with administration of justice,
In a situation where there is close proximity between the defendants house and the witness or if the
defendant has easy access to the witness. He can easily go and intimidate the witness or persuade him through
other means not to testify,
The defendant can easily access the evidence. Take a scenario the defendant is accused of stealing a car
or killing a little boy and in both cases the stolen car or the dead boy is not yet traced. Chances are that if
released he may go ahead to ensure that the car is not found or the dead boys body is not discovered,
45
Patricia Hassett,, A Prototype Expert System For Making Bail Recommendations, (1992) 7th BILETA
Conference, Information Technology and Legal Education: Towards 2000, 9th & 10th April 1992
46
Neil Corre and David Wolchover, Bail in Criminal Proceedings, (3rd Edition, Oxford University Press, 2004)
Chapter 1: Right to bail

If there is believe that the defendant may have access to jurors and may bribe them or intimidate them,
the defendant is part of larger group and other suspects are still at large. If released on bail he may tip-off other
suspects.
The defendant has directly threatened the witness or has somehow (be it directly or indirectly) admitted
to do so.
From the foregoing the risk of interfering with the administration of justice can be summarised as show below:
High: Accused and witness live in the same house; Accused lives very near to witness; Accused can
easily access evidence if released; There is a strong likelihood that the accused may bribe or intimidate the
jurors if released; The accused has threatened the witness
Low: Accused is not related to witness; Accused has no access to witness; Accused if released will
have no access to evidence; There is no chance that the accused will intimidate or bribe the jurors; Accused has
in no way threatened the witness
4.2 Design of A BDSS

In building the BDSS an interface would be developed for entering new bail cases; the outcome and the
reasoning behind the bail decision. A list of possible outcomes would be built into the system based on the
analysis under the risk assessment section, see above. Also a list of reasoning behind the outcome will also be
built into the system. Then there will be fields for the decision maker to enter additional comments under the
case outcome field and the reasoning field. For the system to be effective and efficient there must be a large
number of previous cases, outcomes and reasoning behind the outcome entered into the database of the system.
This system will be implemented using web based technology such as ColdFusion or PHP to allow for
centralized access by all judges. The system would also be flexible enough to connect to third party databases
for references.
The system is modelled using the three risk analysis scenarios detailed above. Figure 1 below shows
the process flow diagram for the risk of reoffending. Each of the risk variables is assigned specific weight as
would have been determined by the judicial system, meaning it is basically a hybrid of case-based and rule-
based reasoning structure. The system computes the cumulative weight by multiplying the weight of the variable
by 1 if it is a yes or by 0 if it is a No. The same thing is repeated in figure 2 for the risk of non-appearance and
figure 3 for the risk of interfering with the administration of justice. The points obtained in each of these three
sections are further added together and weighed against set thresholds, see figure 4. If the point gained is above
a given threshold the result will be bail denied, if it is however below the threshold it is bail granted. The bail
granted could be conditional or unconditional depending on how much the point is below the set threshold, see
the scenario under artificial neural networks.

Risk of Reoffending Each risk factor is assigned a value of

1 if answer is YES and 0 if answer is NO. Each risk factor has
a weight which would be aggregated at the end of this
process.
The system
starts checking
each factor
Yes Yes
No 0 * Weighted No Is the offence sex 1 * Weighted
0 * Weighted Is the offender 1 * Weighted
value below 18 years? value value related, assault or value
fraud
Does suspect Yes

0 * Weighted No 1 * Weighted
have a fixed
value address? value Total points
gained
0 * Weighted No Yes
Is offence vehicle 1 * Weighted
value related or value
burglary?
Is suspect Yes
0 * Weighted No
employed? 1 * Weighted
value value
No Has suspect
0 * Weighted Yes 1 * Weighted
previously
value breached bail? value
No Has suspect Yes

0 * Weighted gone to jail before 1 * Weighted
value for a crime? value
Has suspected
No waited for more Yes
0 * Weighted than 6 months for 1 * Weighted
value a case to be value
heard?
Figure 1

Risk of non-appearance Each risk factor is assigned a value

of 1 if answer is YES and 0 if answer is NO. Each risk factor
has a weight which would be aggregated at the end of this
process.
The system
starts checking
each factor
Yes
Does the suspect
value have strong value Total points
family ties? gained
0 * Weighted No Yes 1 * Weighted

Is the suspect
value gainfully value
employed?
Has the suspect

lived in the
0 * Weighted No community for a Yes
long time and has 1 * Weighted
value good ties to the value
community?
No Yes
0 * Weighted Has suspect ever
1 * Weighted
breached jail in
value value
the past?
No Is penalty if
convicted not
value custodial? value
No Does suspect Yes 1 * Weighted

0 * Weighted have a past
value criminal record value
No Is suspect Yes
0 * Weighted mentally stable 1 * Weighted
value and of good value
character?
Figure 2

Risk of interfering with administration of justice Each risk

factor is assigned a value of 1 if answer is YES and 0 if
answer is NO. Each risk factor has a weight which would be
aggregated at the end of this process.
The system
starts checking
each factor
Does the Yes

accused and
value witness live in the value
same house?
Does the Yes

accused live very
value near to witness value
Can the accused Yes

0 * Weighted No
easily access 1 * Weighted
value evidence if value
released?
Is there a strong
likelihood that the Yes
0 * Weighted No accused may bribe 1 * Weighted
value or intimidate the value
jurors if released?
No Can the accused

threaten the
value witness? value
Total points
gained
Figure 3

The merger of different risk scenarios.
Total points gained

Total points gained Total points gained from Risk of
from Risk of Re- from Risk of Non- interfering with
offending appearance administration of
justice
Points are
processed for
the different
risk scenarios
Bail denied, however

Is the points gained No
above the minimum
the judge may use his
threshold? power and discretion to
override this
Yes
Bail granted (could be

conditional or
unconditional)
Figure 4
End the process

Neural technology will later be incorporated into this system when the database has grown large
enough to adequately train the neural network, giving rise to a system similar to what is obtained in split-up47.
Artificial Neural Networks (ANNs) are computer programmes designed to function the same way as the human
brain. They are made up of artificial neurons called neurods which are connected to each other by links. These
links have varying weights. When designing a system, the combined weight of all the input neurods is
compared or measured against a pre-determined threshold48.
ANNs have the advantage of being able to extract patterns and detect trends, as in link analysis,
between two or more related or disparate activities. These patterns and trends ordinarily may not be easily
noticeable by human or other computer systems. For instance ANN can be used to detect if there is a link
between a suspect and previous suspects or convicted criminals. Such revelations may be invaluable in reaching
a bail decision by a judge. ANN can also carry out complex computational analysis in real-time and in parallel
with other computational analysis.
Output signal
Input from ANN trained with previous cases providing
current case as contained in the repository outcome for the
case at hand
Figure 1
ANN compares the input signals provided from the current case with the existing cases stored in the
repository. ANN is able to determine the nature of the crime and predict with substantial level of intelligence,
based on the training it has received, the outcome of the case, see figure 1 above. This is achieved by identifying
the correlation (similarity and pattern) between the current case and existing cases in the database. It matches
the case with the existing cases and filters out the cases that most closely match the current case. Based on the
outcome of the previous cases it can then go ahead to predict the outcome of the current case49. Ordinarily
ANNs do not provide explanation as to how it arrived at a given conclusion.
Let us model a scenario involving a 16 year old unemployed boy who is involved in a vehicle-related
crime. This suspect has previously breached bail and has no fixed address.
47
J Zeleznikow and A Stranieri, Split up: an intelligent decision support system which provides advice upon
property division following divorce(1998) IJLIT 6(2):190-213; Oxford University Press
48
Dan Hunter, Commercialising Legal Neural Networks (7 May 1996) JILT 1996 (2)
<http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/1996_2/hunter/> accessed 28 June 2009
49
Marco Costa, Orlando Sousa and Jos Neves, An Architecture to Legal Distributed Case Repositories,
JURIX 1998 <http://www.jurix.nl/pdf/j98-02.pdf> accessed June 26 2009
Below 18 years of
age
(1=yes, 0=No)
0.
65
No-Fixed_Address
(1=yes, 0=No) 0 .8
5
Vehicle- Sum all weighted inputs Bail_Granted

Related_Crime 0.70 and compares against (<2.0= Bail Granted,
(1=yes, 0=No) set threshold 2.0-2.5=Low bail,
>2.5=No Bail)
Breached_Bail 0.80
(1=yes, 0=No)
Unemployed
75
0.
(1=yes,
0=No)
Figure 2
From the case above which is graphically represented in figure 2 above the system can predict the
possible outcome of the case by adding up the input and comparing it against the set threshold.
(1x0.65) + (1x 0.85) + (1x0.70) + (1x0.80) + (1x0.75) = 3.75 This outcome compared to the set threshold will
suggest that no bail should be granted to the suspect. However, if there is some changes to the status of the
suspect the outcome may be significantly different. If the status of the suspect where to be different, say the
offender has a fixed address and is gainfully employed it could change the output. In such a scenario the
resulting outcome will be: (1x0.65) + (0x 0.85) + (1x0.70) + (1x0.80) + (0x0.75) = 2.15
This means that low bail or conditional bail may be granted to the suspect based on the set threshold.
V. CONCLUSION
Decision support system can be used to improve the efficiency of the bail system. This system can
provide substantial support to the decision making process, providing the decision maker with accurate
information in the right format. It can process huge and complex information within a very short period, saving
time and freeing the decision maker to focus on more delicate issues pertaining to bail such as analysing the
systems final output and reaching final conclusion.
Decision support system is only as effective as those who conceived and designed it. To be able to
build a robust system, good time has to be spent on brainstorming and planning. And those who interpret the
outputs from the system should be very knowledgeable and should be able to provide sound feedback for
immediate use and for future upgrade or redesign of the system. It is recommended that legal professionals be
part of the design and implementation of the bail decision system.
Finally, the decision support system as proposed in this paper should be built on open architecture,
should be easily upgradeable, should have a simple interface, built on web technology and easy to use. As the
number of bail cases in the repository grows, neural network technology may then be introduced. These
previous cases will be used to train and retrain the neural network until it is able to take intelligent decision on
its own with minimal or no support from a user. However, the decision reached by the system is subject to
interpretation by the user (in this case the judge). The judge too will require training on how to use the system.
BIBLIOGRAPHY
[1] A.N Doob and N.W Park, 'Computerised Sentencing Information for Judges' (1987) Criminal Law Quarterly 30: 5472.
[2] Adrian Turner, Getting It Right From The Start, Justice of the Peace, 10 August 2002.
[3] Alan Travis, Overcrowding blamed for rise in prison deaths The Guardian (London Saturday 22 September 2007)
<http://www.guardian.co.uk/uk/2007/sep/22/ukcrime.prisonsandprobation> accessed 20 May 2009

[4] Albert Alschuler, Preventive Pre-trial Detention and the Failure of Interest-Balancing Approaches to Due Process, (1986) 85
Michigan Law Review 510.
[5] Andrew Von Hirsch, Sentencing guidelines and penal aims in Minnesota (1994) Criminal Justice Ethics vol. 13, 1994
[6] Andrew Vincent and John Zeleznikow, Towards a plea bargaining decision support system for Legal Aid lawyers in Victorian
lower courts, John Zeleznikow, Arno R. Lodder, Second international ODR Workshop, Tilburg: Wolf Legal Publishers, pp 47-
,58 2005 <http://odrworkshop.info/papers2005/odrworkshop2005VincentZeleznikow.pdf> accessed 27 May 2009
[7] Andrew Greinke, Legal Expert Systems - A Humanistic Critique of Mechanical Legal Inference (1994) MUEJL Volume 1,
Number 4 <http://www.murdoch.edu.au/elaw/issues/v1n4/greinke14.html> accessed 29 March 2009
[8] Arno R. Lodder, Anja Oskamp and Marius J.A. Duker, AI & Criminal Law: Past, Present & Future JURIX 1998
<http://www.jurix.nl/pdf/j98-05.pdf> accessed 30 May 2009
[9] Austin Lovegrove, Statistical Information--Systems as a Means to Consistency and Rationality in Sentencing (1999) IJL&IT
1999 7 (31)
[10] Baroness Scotland, Comment: Closing The Justice Gap, Law Society Gazette, 4 December 2003.
[11] BBC News, Prison overcrowding 'at crisis point' BBC News (London Wednesday, 28 August, 2002)
<http://news.bbc.co.uk/2/hi/uk_news/2222022.stm> accessed 20 May 2009
[12] BBC News Murder case bail ban 'problems' BBC News (London Tuesday, 17 June 2008)
[13] < http://news.bbc.co.uk/1/hi/uk_politics/7458211.stm> accessed 5 June 2009
[14] B.E. Leake, Case-Based Reasoning: Experiences, Lessons, & Future Directions. (AAAI Press, Cambridge, MA, 1996) a
[15] Chan J. 'A Computerised Sentencing Information System for New South Wales Courts' (1991) Computer Law and Practice, 137
150.
[16] Christos Stergiou and Dimitrios Siganos, 'Neural Networks'
<http://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol4/cs11/report.html> accessed 29 May 2009
[17] Criminal Justice System, Upholding the Rights of the Defendants <http://www.cjsonline.gov.uk/defendant/bail/> accessed 18
April 2009
[18] Cyrus Tata, John N. Wilson and Neil Hutton, Representations of Knowledge and Discretionary Decision-Making by Decision-
Support Systems: The Case of Judicial Sentencing (1996) JILT 1996 (2)
<http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/1996_2/tata> accessed 4 June 2009
[19] D. Brown, Offending on Bail and Police use of Conditional Bail(1998) Home Office Research and Statistics Directorate,
Research findings No. 72, London: Home Office, 1998, p.1
[20] Dan Hunter, Commercialising Legal Neural Networks (7 May 1996) JILT 1996 (2)
<http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/1996_2/hunter/> accessed 28 June 2009
[21] Daniel C. Richman The Story of United States v. Salerno: The Constitutionality (2005).
<http://lsr.nellco.org/cgi/viewcontent.cgi?article=1011&context=fordham/flfc> accessed 17 May 2009
[22] David Bainbridge, CASE': Computer Assisted Sentencing in Magistrates' Courts (2005) 5th BILETA Conference
[23] Dorset Police Force, Bail Conditions <http://www.dorset.police.uk/default.aspx?page=1024> accessed 16 May 2009
[24] Emilia Bellucci and John Zeleznikow, Developing Negotiation Decision Support Systems that support mediators: a case study
of the Family_Winner system (2006), Journal of Artificial Intelligence and Law, 13(2): 233-271.
[25] Emilia Bellucci and John Zeleznikow, Representations for decision making support in negotiation (2001) Journal of Decision
Support, volume 10(3-4): 449-479.
[26] Hubert L. Dreyfus and Stuart E. Dreyfus, From Socrates to Expert Systems: The Limits and Dangers of Calculative Rationality.
<http://www.psych.utoronto.ca/users/reingold/courses/ai/cache/Socrates.html> accessed 25 March 2009
[27] Ivan Potas et al, Informing the Discretion: The Sentencing Information System of the Judicial Commission of New South
Wales, (1998), IJL&IT 1998 6 (99)
[28] J. Hogarth, Sentencing as a Human Process (University of Toronto Press: Toronto 1971).
[29] J. Hogarth, Computer and the Law: Sentencing Database System, User's Guide (Vancouver: LIST Corporation 1988)
[30] James J. Pomykalsk,i Walter F. Truszkowski and Donald E. Brown, Expert Systems (1999)
<http://agents.gsfc.nasa.gov/papers/pdf/es.pdf> accessed 5 April 2009>
[31] Jean-Charles Pomerol and Patrick Brzillon, Context Proceduralization in Decision Making. <http://www-
poleia.lip6.fr/~brezil/Pages2/Publications/26800491JCP.pdf> accessed 2 April 2009
[32] Jenne Van Der Vinne, Ing. W Van Zwol and M Karnekamp A Sentencing Information System Named 'NOSTRA' (1998)
IJL&IT 1998 6 (230)
[33] John S. Goldkamp and Michael R. Gottfredson, Bail decision making and pre-trial detention Surfacing judicial policy (2005)
Springer Netherlands, Law and Human Behaviour
[34] J Zeleznikow and A Stranieri, Split up: an intelligent decision support system which provides advice upon property division
following divorce(1998) IJLIT 6(2):190-213; Oxford University Press
[35] Joseph A. Cannataci, Law, liability and expert systems (1989) AI & Society Volume 3, Number 3 / July, [1989] 169-183
[36] Joseph Weizenbaum, Computer Power and Human Reason: From Judgement to Calculation, (W. H. Freeman & Co., New
York, NY, USA, 1976)
[37] J.T. St. Johanser and R.M. Harbridge, Validating Expert systems: Problems and Solutions in Practice (KBS-86, London, 1986)
[38] K. White, From Knowing to Legal Knowledge: Using Early Twentieth Century Canadian Murder Trials to Problematize
Knowledge Management Technology (2004) CAUT Law Forum - Winnipeg, 2004
[39] Kenneth Russell, Electronic Monitoring (1987) New Law Journal, 20 March 1987.
[40] Larry Laudan, Truth, Error, and Criminal Law: An Essay in Legal Epistemology, (Cambridge Studies in Philosophy and Law,
CUP, Cambridge 2005)
[41] Laurie L. Levenson , Detention, material witnesses & the war on terrorism, (2003) <http://llr.lls.edu/volumes/v35-
issue4/levenson.pdf> accessed 28 June 2009
[42] Legal Services Commission, What is bail? (South Australia, Law handbook)
<http://www.lawhandbook.sa.gov.au/ch02s03s01.php> accessed 6 June 2009
[43] Lisa M. Seghetti and Alison M. Smith, CRS Report for Congress, Federal Sentencing Guidelines: Background, Legal Analysis,
and Policy Options Updated June 30, 2007 <http://www.fas.org/sgp/crs/misc/RL32766.pdf> accessed 3 June 2009
[44] M. J. J. Hall et. al., Supporting discretionary decision-making with Information Technology: A study in the criminal sentencing
jurisdiction, (2005) University of Ottawa Law & Technology Journal

[45] Marco Costa, Orlando Sousa and Jos Neves, An Architecture to Legal Distributed Case Repositories, (1998) JURIX
<http://www.jurix.nl/pdf/j98-02.pdf> accessed June 26 2009
[46] Mary T. Phillips, Release and bail conditions in New York (2004) New York City Criminal Justice Agency, Research brief No.
6, August 2004 <http://www.cjareports.org/reports/brief6.pdf> accessed 7 June 2009
[47] Michael White and Andrew Sparrow, Conservatives unveil plan to reform bail system Guardian (London 11 August 2008)
[48] N. Hutton, A. Patterson, C. Tata, and J.N Wilson,. 'Decision Support for Sentencing in a Common Law Jurisdiction' (1995) Fifth
International Conference on Artificial Intelligence and Law (ICAIL-95), .8995. Washington D.C.: ACM Press.
[49] N. Hutton, C. Tata, and J.N Wilson, 'Sentencing and Information Technology: Incidental Reform?' (1996) International Journal
of Law and Information Technology 3(2): 255285.
[50] N. Hutton, A. Paterson, C. Tata, and J.N Wilson, A Sentencing Information System for the Scottish High Court of Justiciary
(Scottish Office Central Research Unit: Edinburgh 1996);
[51] Neil Corre and David Wolchover, Bail in Criminal Proceedings, (3rd Edition, Oxford University Press, 2004) Chapter 1: Right
to bail
[52] Nigel Morris, Rise in prison suicides blamed on overcrowding The Independent (London Friday, 17 June 2005)
<http://www.independent.co.uk/news/uk/crime/rise-in-prison-suicides-blamed-on-overcrowding-494408.html> accessed 19 May
2009
[53] Patricia Hassett, Can Expert Systems Improve the Exercise of Judicial Discretion? 91992) In: Grtters, C.A.F.M., J.A.P.J.
Breuker, H.J. van den Herik, A.H.J. Schmidt and C.N.J. de Vey Mestdagh (eds), Legal Knowledge Based Systems: Information
Technology & Law, JURIX'92, Koninklijke Vermande, Lelystad, NL, 992. <http://www.jurix.nl/pdf/j92-19.pdf > accessed 24
June 2009
[54] Patricia Hassett, A Prototype Expert System For Making Bail Recommendations, (1992) 7th BILETA Conference, Information
Technology and Legal Education: Towards 2000, 9th & 10th April 1992
[55] Patricia Hassett, Can Expert System Technology Contribute to Improved Bail Decisions? (1993) International Journal of Law
and Information Technology. Vol. 1 No. 2 Summer
[56] Patricia Hassett, Problems in Selecting Effective Computer Technology for Use at the Bail Stage of the Criminal Justice
System (1990) Proceeding of the 2nd National Conference on Law, Computers and Artificial Intelligence
[57] Patricia Hassett, An Expert System for Improving the Pre-trial Release/Detention Decision (1991). Pre-Proceedings of the 6th
Annual BILETA Conference
[58] P. Alldridge, "Anoraks among the suits and jeans computers, law and the legal academy", (1997) Journal of Information, Law
and Technology (JILT).
[59] Paralegal Advisory Service (PAS), What is bail? (2004) <www.penalreform.org/resources/bro-2004-
what-is-bail-en.pdf> accessed 16 May 2009
[60] Patricia M Morgan and Paul F Henderson, Remand decisions and offending on bail: evaluation of the Bail Process Project
(London: Home Office, 1998) Home Office Research Studies
[61] <http://www.homeoffice.gov.uk/rds/pdfs/hors184.pdf> accessed 18 May 2009
[62] Peter Wilcock and Joel Bennathan, The new bail system (2004) New Law Journal Issue 7126
[63] Prison Reform Trust, Five ways to stem prison overcrowding, (2005)
<http://www.prisonreformtrust.org.uk/subsection.asp?id=349> accessed 20 May 2009
[64] Robert Jago, Civil and Criminal Procedure
<http://www.londonexternal.ac.uk/current_students/programme_resources/laws/subject_guides/civ_crim/criminal_procedure_ch
11.pdf> accessed 8 June 2009
[65] Samantha Besson, Four Arguments Against Compromising Justice Internally, (2003) Oxford Journal of Legal Studies, 1 June
2003
[66] Schmatt E. 'Judicial Information Research System (JIRS)' (1996) Judicial Officers Bulletin 8(7): 4950. Judicial Commission of
New South Wales.
[67] Sharon Hanson, Legal method and reasoning, (2nd Edition Cavendish Pub Ltd, London 2003)
[68] Sophie Goodchild, Public at risk from prison overcrowding The Independent (London Sunday, 21 January 200)
< http://www.independent.co.uk/news/uk/crime/public-at-risk-from-prison-overcrowding-433075.html> accessed 20 May 2009
[69] Stephen Jones , Guilty until Proved Innocent? The Diminished Status of Suspects at the Point of Remand and as Unconvicted
Prisoners, (2003) Common Law World Review, 1 December 2003.
[70] Steve Chanenson, Marc Miller and Ron Wright Users and Uses of Sentencing Data 92006) NASC Annual Conference
<http://www.ussc.gov/states/2006conf/Chanenson%20Miller%20Wright%20Tata%20Hutton%20Plenary%20 -
%20Policy%20Shaping.pdf> accessed 1 June 2009
[71] Susan Edwards, What shall we do with a frightened witness? (1989) New Law Journal, 22 December 1989.
[72] The Sentencing Commission for Scotland, The Scope to Improve Consistency in Sentencing Report 2006
<http://www.scottishsentencingcommission.gov.uk/docs/consistency/Consistency%20Report%20-%20Final.pdf> accessed 30
May 2009
[73] The Law Commission, Bail and the Human Right Act 1998: Executive Summary LawCom No 269,
<http://www.lawcom.gov.uk/docs/lc269sum.pdf> accessed 5 June 2009
[74] The Law Commission, Criminal Law: Bail and The Human Rights Act 1998: A Summary (1999) Consultation Paper NO 157
<http://www.lawcom.gov.uk/docs/cp157sum.pdf> accessed 17 May 2009
[75] The Journal Online, Bail system in crisis, say police chiefs: Conference call to Executive for crackdown on bail offenders The
Journal Online (London 20 May 05)
<http://www.journalonline.co.uk/News/1001851.aspx> accessed 4 July 2009
[76] Thomas F. Gordon, Henry Prakken and Douglas Walton The Carneades Model of Argument and Burden of Proof (2007)
<http://www.dougwalton.ca/papers%20in%20pdf/07GordonPrakkenWalton.pdf> accessed 4 August 2009
[77] U.J Schild, Criminal sentencing and intelligent decision support, (1998) 6 Artificial Intelligence and Law 151.
[78] U.J Schild and J. Zeleznikow, A comparative study of decision support systems for the sentencing of criminals (2008) Journal
of Decision Systems, 17 (4):523-552.
[79] United States Sentencing Commission, Federal Sentencing Guidelines Manuals (2008)
<http://www.ussc.gov/2008guid/GL2008.pdf > accessed 3 June 2009
[80] Y. Hacohen-Kerner and U.J. Schild, Case-based Sentencing Using a Tree of Legal Concepts, (2001) Information &
Communications Technology Law, Volume 10, Issue 1 March 2001 , pages 125 - 135

Cases
Lord Russell in R v Rose (1898) 78 LT 119
Linford v Fitzroy (1849) 13 QB 240 at 247;
R v Badger (1843) 4 QB 468 at 472 and Osborne V Gough 3 B & P 551.
R v Vallet [1951] 1 All ER 231
R(O) v Harrow Crown Court (2003) The Times, 29 May by Kennedy LJ
U.S. v. Booker, 543 U.S. 220 (2005)

Bail Decision Support System PDF

Uploaded by

Copyright:

Available Formats

Bail Decision Support System PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bail Decision Support System PDF

Uploaded by

Copyright:

Available Formats

The International Journal Of Engineering And Science (IJES)

|| Volume || 3 || Issue || 8 || Pages || 45-66 || 2014 ||

Bail Decision Support System

II. OVERVIEW OF BAIL SYSTEM

www.theijes.com The IJES Page 46

2.1 Breach of Bail

2.2 Challenges of the Present bail System

www.theijes.com The IJES Page 47

www.theijes.com The IJES Page 48

www.theijes.com The IJES Page 49

2.2.1 Problems with the Bail Act

www.theijes.com The IJES Page 50

III. SENTENCING DECISION SUPPORT SYSTEMS

www.theijes.com The IJES Page 51

ENGLAND AND WALES

www.theijes.com The IJES Page 52

NEW SOUTH WALES, AUSTRALIA

www.theijes.com The IJES Page 53

IV. BAIL DECISION SUPPORT SYSTEM

4.1 Risk Assessment: Reasons for Developing a BDSS

www.theijes.com The IJES Page 54

4.1.1 Risk of Reoffending

www.theijes.com The IJES Page 55

4.1.2 Risk of non-appearance

4.1.3 Risk of interfering with the administration of justice

www.theijes.com The IJES Page 56

4.2 Design of A BDSS

www.theijes.com The IJES Page 57

Risk of Reoffending Each risk factor is assigned a value of

Does suspect Yes

No Has suspect Yes

www.theijes.com The IJES Page 58

Risk of non-appearance Each risk factor is assigned a value

0 * Weighted No Yes 1 * Weighted

Has the suspect

No Does suspect Yes 1 * Weighted

www.theijes.com The IJES Page 59

Risk of interfering with administration of justice Each risk

Does the Yes

Does the Yes

Can the accused Yes

No Can the accused

www.theijes.com The IJES Page 60

The merger of different risk scenarios.

Total points gained

Bail denied, however

Bail granted (could be

www.theijes.com The IJES Page 61

Vehicle- Sum all weighted inputs Bail_Granted

www.theijes.com The IJES Page 63

www.theijes.com The IJES Page 64

www.theijes.com The IJES Page 65

www.theijes.com The IJES Page 66

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.