Network Analysis with Stochastic Grammars
Abstract
Digital forensics requires significant manual effort to identify items of evidentiary interest from the ever-increasing volume of data in modern computing systems. One of the tasks digital forensic examiners conduct is mentally extracting and constructing insights from un structured sequences of events. This research assists examiners with the association and individualization analysis processes that make up this task with the development of a Stochastic Context -Free Grammars (SCFG) knowledge representation for digital forensics analysis of computer network traffic. SCFG is leveraged to provide context to the low-level data collected as evidence and to build behavior profiles. Upon discovering patterns, the analyst can begin the association or individualization process to answer criminal investigative questions. Three contributions resulted from this research. First , domain characteristics suitable for SCFG representation were identified and a step -by- step approach to adapt SCFG to novel domains was developed. Second, a novel iterative graph-based method of identifying similarities in context-free grammars was developed to compare behavior patterns represented as grammars. Finally, the SCFG capabilities were demonstrated in performing association and individualization in reducing the suspect pool and reducing the volume of evidence to examine in a computer network traffic analysis use case .
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 17, 2015
- Accession Number
- ADA621776
Entities
People
- Alan C. Lin
Organizations
- Air Force Institute of Technology