Greetings,<br><br>I am pleased to announce the call for papers for a
special issue of Security Informatics: Fusing Automatic Text Processing
with Criminal Incident Data. A brief synopsis is given below. Please
see <a href="http://ptl.sys.virginia.edu/msg8u/cfp_final.pdf">this link</a> (PDF) for the submission schedule and additional
information.<br>
<br>=====================================<br><div id=":l1">Crime analysts often use an area’s historical record to visualize past crimes (e.g., using hot-spot mapping)<br>and to predict locations of future criminal activity. Models for the latter task use geographic and demographic<br>
factors to characterize the appeal of potential crime sites, demonstrating promising performance on real-world<br>prediction tasks (Fox et al., 2012; Wang and Brown, 2012). However, these models often ignore the vast<br>
repository of unstructured text that is freely available through, for example, news and social media outlets.<br>
Such information sources contain detailed descriptions of past, present, and future events, and recent work<br>has shown that these descriptions can improve crime prediction performance (Wang et al., 2012). Despite<br>this encouraging result, textual information remains largely unexploited due to its vast size and unstructured<br>
format. This special issue of Security Informatics will focus on fusing text processing outputs (e.g., events, facts,<br>and opinions) with historical criminal incident data (e.g., spatio-temporal criminal incident locations). Such<br>
work will help bridge the current gap between unstructured text and crime analytics (e.g., predictive policing).<br><br>In particular, we welcome high-quality submissions on the following topics:<br>* Extraction and geocoding (address resolution) of event locations within unstructured text<br>
* Extraction and normalization of event times within unstructured text<br>* Extraction of person/group names and sentiment from unstructured text<br>* Processing of “noisy” sources of unstructured text (e.g., Twitter and weblogs)<br>
* Fusion of the above (or other) textual information with criminal incident data<br>=====================================<br><br>Sincerely,<br><br>Matthew Gerber<br>Lead Guest Editor<br>Department of Systems and Information Engineering<br>
University of Virginia</div>