Graph Analytics for Signature Discovery

Conference · Sat Jun 01 00:00:00 EDT 2013

DOI:https://doi.org/10.1109/ISI.2013.6578850· OSTI ID:1144838

Hogan, Emilie A.; Johnson, John R.; Halappanavar, Mahantesh; Lo, Chaomei

Within large amounts of seemingly unstructured data it can be diffcult to find signatures of events. In our work we transform unstructured data into a graph representation. By doing this we expose underlying structure in the data and can take advantage of existing graph analytics capabilities, as well as develop new capabilities. Currently we focus on applications in cybersecurity and communication domains. Within cybersecurity we aim to find signatures for perpetrators using the pass-the-hash attack, and in communications we look for emails or phone calls going up or down a chain of command. In both of these areas, and in many others, the signature we look for is a path with certain temporal properties. In this paper we discuss our methodology for finding these temporal paths within large graphs.

• Research Organization: Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC05-76RL01830
• OSTI ID: 1144838
• Report Number(s): PNNL-SA-94756
• Resource Relation: Conference: IEEE International Conference on Intelligence and Security Informatics (ISI 2013), June 4-7, 2013, Seattle, Washington, 315-320
• Country of Publication: United States
• Language: English

Related Subjects

Signature Science