Exploiting Essential Elements of Information from Significant Activity Reports (SIGACTS) for Forensic Analysis

The goal of this research is to provide automated information extraction services, in particular relationship extraction, for forensic analysis. Specifically, we propose to develop and pursue an ontology-driven approach to information extraction, in which a formal ontology capturing concepts of interest is used not only to facilitate the exploitation of extraction results in a larger forensic analysis system but also to drive the development of the primitive language resources and pattern recognizers that provide those extractions services at the lowest level. A formal standards compliant ontology facilitates communication between developers and other stakeholders, can serve as a specification for new language resources required by the SIGACT domain, makes it easier for users to understand and evaluate the system, makes the results more useful to downstream applications, and facilitates the integration of the IE engine into a service oriented application environment. The focus will be on the characterization and extraction of relationships of interest, as these allow new visualization tools (network visualization) to be used for forensic analysis and also serve as a potential bridge to new items of related data for the analyst''s review. BENEFIT: This research will make it easier to extend our information extraction capabilities to new domains and should improve our ability to specify and engineer individual linguistic knowledge resources. It should also allow our information extraction products to be more easily integrated into application environments emphasizing collaborating services.