Transcript An Intelligence Proccess-driven Knowledge Extraction Framework for

New Challenges in the European Area
Young Scientist's 1st International Baku Forum
May 20-25
PhD Thesis – Research Plan
ALBERTETTI Fabrizio
Thesis Director: Prof. STOFFEL Kilian
Information Management Institute
University of Neuchatel
Switzerland
Context
Objectives
Research
Challenges
2
» Interdisciplinary project:
˃ Computational
+ Information Management Institute, University of
Neuchatel
˃ Forensics
+ Institut de Police Scientifique, University of Lausanne
» Supported by the Swiss National Science Foundation (SNSF)
» 5 years project (?) – Started in Sept. 2011
3
4
» E.g., the routine activity approach
(Cohen & Felson, 1979)
Figure: Routine Activity (popcenter.org)
5
"Crime analysis is the systematic study of
crime and disorder problems as well as other
police-related issues—including
sociodemographic, spatial, and temporal
factors—to assist the police in criminal
apprehension, crime and disorder reduction,
crime prevention, and evaluation."
(Boba, 2005)
6
"Crime analysis is the systematic study of
crime and disorder problems as well as other
police-related issues—including
sociodemographic, spatial, and temporal
factors—to assist the police in criminal
apprehension, crime and disorder reduction,
crime prevention, and evaluation."
(Boba, 2005)
7
» The chain of events in crime prevention:
Prevention
Proactivity
Predictability
Patterns
From patterns to prevention (Ratcliffe, 2009)
Computational
Forensics !

Discovering Forensic
Knowledge
8
» To develop a framework :
• For conducting analyses
• Driven by processes (using domain
knowledge)
• Intelligent (assessing the results)
• Extracting knowledge from forensic data
9
» What is the nature of forensic data?
˃ Uncertain
˃ Incomplete
˃ Inaccurate
» Why?
˃ Because it is based on hypotheses and conjectures
˃ Because it stems mainly from latent marks
˃ Because it reflects the effects and not the causes
(abduction)
10
Challenges:
» To conduct analyses and perform
deduction/reasoning with partial knowledge,
uncertainties and conjectures
» To integrate domain intelligence for providing
practical and consistent results
» To conduct analyses with a holistic view of the
macro process, i.e. combining several mining
outcomes based on crime analysis processes
11
DOMAINDRIVEN
DATA
MINING
KNOWLEDGE
REPRESENTATION
COMPUTATIONAL
FORENSIC
FRAMEWORK
FUZZY
LOGIC
FORENSIC
SCIENCE
12
» Computational forensics is still an
emerging research area
» Only a combination of several domains
can answer crime analysis questions
13
PhD Thesis – Research Plan
ALBERTETTI Fabrizio
Thesis Director: Prof. STOFFEL Kilian
Information Management Institute
University of Neuchatel
Switzerland
* This project is supported by the Swiss National Science Foundation