Geography Reference
In-Depth Information
may be an imperfect fit for a later entry of a different crime. Therefore a new
category is created that fits that later entry but overlaps semantically with the
earlier-created category (e.g. ''organised crime'' and ''gang dispute'' as reasons for
a crime). This phenomenon creates issues when trying to glean patterns using the
eXplorer tool - the imperfect classification may be making patterns less apparent
or worse, invisible. The use of categorized crime with close meanings (e.g. robbery
and theft) and, at a higher level, the semantic closeness of crime type and crime
victim type only adds to the uncertainty. Elwood (
2009
) acknowledges both this
heterogeneity due to 'diverse categorization schemes' and the shifting nature of a
dataset such as WikiCrimes' as part and parcel of volunteered data.
Another issue is the trustworthiness of the data itself, as it does not come from
an authoritative source and is open to abuse. However, the amount of reports that
come in and the fact that for this study the reports are aggregated means that the
conclusions derived are based on more robust data than an individual report may
be (i.e. there is safety in numbers).
Finally, there is a possible bias towards states with more reports contained
within its boundaries. Despite the efforts at normalization from counts to per-
centage proportions, for states such as Ceará in particular, having the lion's share
of reports, bias will manifest itself in having a greater variety of reasons for crime,
crime settings and even crime types and crime victim types. The sheer amount of
scenarios for reported crime in Ceará and particularly in Fortaleza, and the
potential variety it brings, make this more statistically likely.
Future directions include the linked analysis of this volunteered data with
census data, which could potentially yield more insights, especially at the (Fo-
ratleza) city scale, for which most of the data exists. Other datasets could be used
to link with the volunteered crime data, for example car ownership rates or tourism
data (as a group of people least likely to make a wiki entry following a crime—is
this a source of silence in the data?) Use of eXplorer's visual time analysis tools
could be used to mine temporal trends from data segmented into intervals.
Acknowledgements The authors would like to thank Tobias Åström at Norrköping Commu-
nicative Visual Analytics (NComVA) in Sweden for providing access to Open Statistics
eXplorer. The feedback from the two reviewers of this chapter is also appreciated.
References
Andrienko G, Andrienko N, Jankowski P, Keim D, Kraak M-J, MacEachren A, Wrobel S (2007)
Geovisual analytics for spatial decision support: setting the research agenda. Int J Geogr
Inform Sci 21(8):839-857
Chainey S, Ratcliffe J (2005) GIS and crime mapping. Wiley, Chichester
Elwood S (2009) Geographic information science: new geovisualization technologies—emerging
questions and linkages with GIScience research. Prog Hum Geogr 33(2):256-263
Furtado V, Ayres L, de Oliveira M, Vasconcelos E, Caminha C, D'Orleans J, Belchior M (2010)
Collective intelligence in law enforcement—the WikiCrimes system. Inf Sci 180(1):4-17
