Page 106 - Proceedings of the 2017 ITU Kaleidoscope

2017 ITU Kaleidoscope Academic Conference

2. CRIME CONTROL IN SOUTH AFRICA internet-crawling to extract crime data that will assist real-
time information capture, as well as automated and timely
The South African Police Service (SAPS) is the national po- documentation of crime information.
lice force of the Republic of South Africa, with about 1,138
police stations. These stations are divided according to the 2.2.2. Limitation on Knowledge Acquisition
9 provincial borders, and a Provincial Commissioner is ap-
pointed in each province [2]. A major responsibility stipu- The usefulness of the information supplied by the current
3
lated by the country’s constitution is that SAPS should pre- system (e.g. Crime Hub ) to stakeholders and community
vent, combat and investigate crime. However, containing policing authorities is rather wanting. The summary crime
crime remains a never-ending concern in South Africa (SA), statistics usually reported by SAPS are at best only able to
1
as suggested by the victim of crime survey . In recent times, provide a rough indication of deterioration or improvement
there has been a significant increase in the crime rate in South within different suburbs in South Africa or between police
Africa, and this has been a major motivation for this research. districts. It would for example be more desirable to report
on exact attributes or the peculiar (characterising) features of
a crime trend, which can assist actionable knowledge sup-
2.1. Crime Control - Current Practice port. For example, it is not sufficient to report that; “1700
sexual assaults cases were recorded in western province in
In SA, crime reporting is typically done at the police sta- 2016”; rather, a report stating that, “out of 1700 sexual as-
tion using traditional (manual) approach. This information is saults cases recorded in western province in 2016, 500 of
later captured onto the system at the local police stations and the cases have been identified to involve repeat offender(s),
stored for (future) processing at a provincial level, where a who mostly operates at night (between 7-9 pm) at the city-
domain expert analyses the information for knowledge sup- centre and captures young females between ages 12 to 25 as
port. This is typically the case because existing software (e.g. victims”. The latter report reveals more information about
Analyst’s Notebook) that could reveal pattern in crime data the spatial, nature and sensitivity of crime attributes involved
are very expensive to purchase and requires critical training in such series. Hence, can aid actionable knowledge (e.g.
or a domain expert, which poses serious constraints on de- suspect prioritisation) for crime deterrence in resource con-
veloping nations. Our findings indicate that such tool is only strained settings.
available at the headquarters/provincial level in SA. Thus,
local stations only make use of basic Excel software for fil-
2.2.3. Potential Delay in Crime Information Dissemination
tering data and identifying patterns, which is cumbersome,
error prone and time consuming. This is a great hindrance to
Considering the current practice in SA, there is high tendency
effective policing.
for delay in crime mitigation practices, leading to poor in-
terventions and policing strategies. This is evident as local
stations typically transfer crime data accumulated over a pe-
2.2. Crime Control - The Gaps
riod of time to the provincial authority for analysis, since
2
According to report , the South African police:citizen ratio there are few domain experts that can handle such analysis
and that is where a more advanced tool is available. This is a
is currently 1:347; that is one police officer for every 347 cit-
great limitation to effective policing because if at local levels,
izens, or around 288 police officers per 100,000 people. This
police are able to derive patterns in a timeous manner, then
report positions the country in the lower-middle end of polic-
they can act to stop such patterns. However, in situations
ing when compared to countries across the world. While the
where they will have to wait a couple of days or weeks to get
police is determined to “squeeze crime to zero” [2], there
the analysed pattern from provincial level, crime could have
exists some challenges that may hinder their effort in com-
worsened during the waiting period. This gap can be fixed
bating crime. The following four main gaps were identified
by deploying at local levels cost-effective user-centred tools
in the current approach to crime control:
(e.g CriClust), which can present understandable structures,
patterns or trends to stakeholders in a timeous manner.
2.2.1. Limitation on Crime Data Acquisition
2.2.4. Non-Proximity Centred Analysis
The crime data provided by SAPS may suffer from omis-
sion and inaccuracies since it is still manually captured by The aim of data analytics is to transform data into “smart
the police. These inaccuracies could hinder crime mitiga- statistics” (i.e. non-trivial and useful information) to gain in-
tion measures. A related study (by the authors) aims to ad- sight for knowledge support. Crime patterns often differ and
dress some of these challenges by use of a context-aware ap- have their unique Modus Operandi (MO), since the oppor-
plication that integrates crowd-sourcing, mobile phones and tunities available to potential offenders vary across different
spatial space due to differences in spatial factors [5]. Hence,
1 http://www.statssa.gov.za/publications/P0341/ a spatial framework with features and instances embedded
2
https://businesstech.co.za/news/general/95069/south-africas-police-
force-vs-the-world/ 3 https://www.issafrica.org/crimehub/

– 90 –