Computational Framework for Identifying Suspects in Multiple Situations
Student Classification
Senior
Faculty Mentor
Albert Esterline, Ph.D.
Department
Department of Criminal Justice/Department of Computer Science
Document Type
Poster
Publication Date
Fall 2019
Disciplines
Computer Sciences | Criminology and Criminal Justice
Abstract
We present a framework for identity that addresses how information can be evidence for identity hypotheses and how such evidence can be discounted and combined. The framework (which is computational) is built on three pillars: the situation theory of Barwise and Perry (and Devlin), the Dempster-Shafer theory of evidence, and Semantic Web standards (OWL, RDF, and the use of URIs). According to situation theory, situations support information and some (particularly utterance situations) carry information about other situations. We see a legal case investigating the identity of an agent as a constellation of situations, which provide evidence for identity hypotheses. We have developed OWL ontologies to provide concepts for encoding cases in RDF. The structure captured in these encodings allow us to apply Dempster-Shafer theory in novel ways to discount and combine levels of evidence for various hypotheses. The majority of this poster is an in-depth analysis of our new scenarios that have multiple situations.
Recommended Citation
Foster, Hannah, "Computational Framework for Identifying Suspects in Multiple Situations" (2019). Undergraduate Research and Creative Inquiry Symposia. 103.
https://digital.library.ncat.edu/ugresearchsymposia/103