Ego-Centric Approach For Predicting Fraudulent Collaboration In Telecommunication
Recently, there has been a surge of interest in social networks ever since the tragic event of September 11, 2001 attacks on The World Trade Center in the United States. E-mail traffic, disease transmission, criminal activity and communication network can all be modeled as social networks. Ego-ce...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2010
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/11933/1/FSKTM_2010_1.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Recently, there has been a surge of interest in social networks ever since the tragic
event of September 11, 2001 attacks on The World Trade Center in the United
States. E-mail traffic, disease transmission, criminal activity and communication
network can all be modeled as social networks. Ego-centric is an approach used in
social network analysis. In the social network parlance, the focused person is referred
to as “ego” and his or her affiliate, friend or relative is known as “alters”. An egocentered
network positions an individual at the center of a social network team for
the person to traverse his or her relationships with other team members. Through
social network analysis, enforcement officers can recognize how information flows
through social ties, how people acquire information and resources and how cleavages
and coalitions operate. In this thesis, based on social network theories and link
analysis; a data mining technology, a social network analysis model is developed to
facilitate in detecting fraudulent collaboration, after which an evaluation is then
made on the performance of the developed model. This study aims to explore the usage of embedding social network analysis functions into fraudulent collaboration
investigation in call details records. Two types of social network data collection
approaches are discussed; (i) social network with centrality measures values and (ii)
social network without centrality measures values, where the first approach is based
on the previous research while the second is based on the current research
experimented. Performance of the models produced by both approaches are
measured based on a standard measurement. Performance is tested using statistical
models which include Bayesian Network, Naïve Bayesian and Binary Logistic
Regression Model is performed. These statistical models are used in order to prove
and determine which model is the ‘best’ that can produce a better prediction of
fraudulent collaboration. The outcome of this research is thought to be of help to any
enforcement agency or relevant authority in its future operations or measures to
detect fraudulent activity in social networks. |
|---|