09/12/2014

Terrorism Historical Database GIS Project - Final Report

2012-09-19

Terrorism Database GIS Project
- Final Report -
by René Brauer

Project start:      09/2011
Project end:       09/2012
Work hours:       96 h

Note: The project was initially thought only to last 3 months; however halfway through the project Fridlund and Brauer started an individual separate course in Science and Technology Studies together. In order to avoid conflict of interested it was decided to postpone the project. The remaining time of the project was utilized sporadically in the year 2012.  

Objectives:

·         Exploring GIS analytical possibilities for historical study of terrorism.
·         Exploring Similar GIS approaches to other historical subjects.
·         Exploring the possibilities of interactive GIS


The basic idea behind the project is the realization that every action is happening at a defined point in time and place and that no action occurs without context. An example of context to the spatiality of terrorist attacks is given by Fridlund (2012)[1] when he writes: “…when Russian idealists discovered the potential for a new form of revolutionary propaganda afforded by small and powerful revolvers.” (Fridlund 2012: 7). This action of ‘discovering’ implies a temporal dimension; dividing a timeline into before the discovery and after. Furthermore it also implies that the ‘discovery’ was transferred to fellow revolutionists, implying a spatial process. These pattern underlying terrorist attacks are difficult to map, since they are so far not well documented, because they are often regarded as criminal activity. By mapping the actual terrorist attacks themselves the project aimed at creating a proxy variable, for the spread of terrorism techniques. In order to analyze the temporal inception and tempo-spatial diffusion of techniques used for assassination by terrorist groups. Utilizing geographical analysis with Geographical Information System (GIS), which are computer aided mapping and analyzing tools.

Spatial Analysis of 19th century Terrorism

Any geographical analysis requires data to analyze; in this case no such geographical data existed in ready analyzable format. Therefore a large part of the project was spent on putting together a database; compiling data into a standardized usable format from different data sources. These data sources varied greatly in their format, ranging from already existing databases, to mentioning’s in books, on internet sites, newspaper clippings etc., varying greatly in amount of data quality.

A great difficulty for the project was the definition of its categories, for example the outcome of a terrorist attack was divided into three different possibilities: fulfilled, failed or foiled. Nevertheless even these three simple categories proved difficult to determine in cases, because if the aim of the terrorist attack was to frighten the target and not kill, than the attack was a success even if the target wasn’t killed. However this presupposes knowledge about the aim of the terrorist’s, this knowledge was in most cases not available. Furthermore if the terrorist attack was foiled, what location should be given, the location of the apprehension or if available the location for the planned attack. These are just a few examples of the difficulties involved in classifying the terrorist attacks, nevertheless most of these were solved by some means or another leading to creation of database including 104 separate terrorist attacks evenly spaced throughout the 19th century. The categories and the subsequent information that the database provided consisted of the following categories:

·         Date                        - date of the terrorist attack
·         Location                 - location of the terrorist attack
·         Perpetrator              - individual who conducted the attack
·         Title of the Target   - if available, royal title or political office  
·         Target                     - name of the victim
·         Method                   - specification over what kind of method
·         Outcome                 - success, failure or apprehension of the attack

Based on the established database general trends were recognized. However the generality of these trends is highly questionable due to the quality of the data.  For example sometimes the only mentioning of method in an information source was: “shot” revealing no information by what method other than some type of firearm. Therefore many entries had to be omitted, leading to only 32 data entries usable for a method analyses (out of a total of 106). Figure 1 shows one type of analyzes that was created from the data material, namely the cumulative occurrence of terrorist attacks utilizing revolvers or other types of weapons. 


Figure 1 crosses (x) represent ‘other types of methods’ while plus signs (+) represent ‘revolvers’

As seen in Figure 1 the usage of revolvers drastically increased around 1880 completely dominating all other methods. However regarding the small sample size of 32 cases it is difficult to assess if this is a general trend or only valid for this particular dataset of 32 of 106 (for this study) known and an unknown total of all terrorist attack in the 19th century.

Therefore the conclusion of the project is that it is intrinsically possible to do this type of analyze possible yielding new insights into the study of historical terrorism. However the limited timeframe for the project did not allow for the creation of a larger sample size that is necessary to generalize any sort of result. 

Possibilities to extend data quality

One idea that was tested in the scope of the project to get around the problem of too few data points was to ‘crowd source’ the problem. I.E. creating a public accessible platform where users would be enrolled in contributing in compiling information for the researcher effectively outsourcing the largest part of the research project, data gathering. As a pilot test a trial was conducted data mining the London Times online newspaper archive. The time span for the search defined itself to the 19th century (1800 – 1900), a search for the keyword ‘murder’ revealed 56 257 articles mentioning ‘murder’ in some form or another. Overwhelmed with this plethora of data a pilot-test was conducted randomly selecting 20 articles and trying to compile their data.  The 20 selected articles were used to create the same data points as mentioned above; date, location etc. The result of this trial was that the London Times as a single data source was not enough to provide all the necessary information to compile the needed data. Further research showed that the data was often available from other sources, libraries, articles websites etc. However this drastically increased the difficulty in the compilation process making it far too complicated for a simple user interface. Furthermore the data itself faced the same problems of ambiguity as the terrorist data explained above, creating possible difficulties in the description. Therefore the collection method might be suitable for a PhD research project, but not for a crowd sourcing project as it presents far too many complexities to one particular user. 

Similar projects, software’s and conclusion of the project

Several GIS approaches have made their way into historical studies, creating se so called subfield of Historical GIS (HGIS). Examples of this kind of study are inquiries about the Salem Witch Trials[2], Battle of Gettysburg[3] and the steel industry of Pennsylvania[4].

Salem Witch Trials:       Reexamination of the trials by mapping the occurrences and the households brought into questions the previously assumed hypothesis about an east west separation. Furthermore analysis on regional scale showed the viral like spread of trials throughout the surrounding areas


Battle of Gettysburg:     The creation of a viewshed (possible points of view) from General Lee´s position revealed that he was unable to physically see the battle firsthand. Therefore he had to rely on second hand information possibly influencing the outcome negatively for the confederate troops.

Pennsylvania iron smelting:        Previous assumptions about the rationality for adopting continental smelting methods due to their superiority of production were called into question. Based on comparisons of actual production data between charcoal and fossil coal, and based on analysis of the locations of the mills and the furnaces.

The main software used behind most of HGIS project represents ArcGis[5], as it is one of the largest GIS software providers. However open source alternatives exist such as TimeMap[6] and QuantumGIS[7] (QGIS). Both are desktop type GIS software’s and very suitable for light type of spatial analysis required for the purpose described here. Quantum GIS may represent the more attractive software option. Firstly it had the possibilities for GRASS[8] plug-ins, opening the possibility for advanced GIS analysis if need. Secondly QGIS has a large community, ensuring continuously updating and problem solving of possible software bugs. Finally QGIS is an Open Source engine also employing Open Source data formats that are compatible with most online services for further online usage of the data.  

The possibility to display the collected data in open source format available for the public was deemed highly possible from the information gathered. Projects like Pelagios[9] specialize in sharing historical humanities research data and making it available, all using Open License data formats. Google Earth[10] allows for similar possibilities and has the advantage of being widely known, however Goggle uses its own data format possibly creating compatibility and licensing issues. Therefore the choice of which online platform to use to make the data available is highly depended on what intentions the creator has with the data. If the only goal is to visualize the data and make this visualization available Goggle Earth is the perfect platform, however if also the core data should be able available for further analysis by third parties Goggle might not be the perfect option.   

The overarching lesson learned from this project:

This type of GIS analysis and distribution is potentially possible, however due to the novelty of this approach large amount of time has to be spend on data collection, analysis and preparing the data before any kind of general results or possible publication can be achieved.



[1] Fridlund, Mats (2012) Affording terrorism: Idealists and materialities in the emergence of modern terrorism, in Terrorism and affordance. ed. Max Taylor & P.M. Currie, London, Continuum
[2] http://www2.iath.virginia.edu/salem/maps.html
[3] http://www.nytimes.com/interactive/2011/07/27/arts/spatial-maps.html
[4] Knowles, A.K and Healy, R. G. (2006). Geography, Timing, and Technology: A GIS-Based Analysis of Pennsylvania's Iron Industry, 1825-1875, Journal of Economic History Vol. 66. Is. p. 608
[5] http://www.esri.com/software/arcgis
[6] http://www.timemap.net/
[7] http://www.qgis.org/
[8] http://grass.fbk.eu/
[9] http://pelagios-project.blogspot.se/
[10] http://www.google.com/earth/index.html

No comments:

Post a Comment