Automatically Tracing Information Flow of Vulnerability and Cyber-Attack Information through Text Strings
Abstract
Quick dissemination of information about new vulnerabilities and attacks is essential to time-critical handling of threats in information security, but little systematic tracking has been done of it. We are developing data mining techniques to track the flow of such information by comparing important information-security Web sites, alert messages, and strings in packets to find similar words and sentences. We report on tools we have developed to collect relevant sentences, with a particular focus on comparing sentences from different sources to find patterns of quotation and influence. We report results on some representative pages that indicate some surprising information flows, for which the combination of both word matching and structure matching performed significantly better than either alone. We also report on preliminary work on the front lines of cyber-attack, trying to correlate text in intrusion-detection reports and even attack packets observed on a honeypot with reports of known attacks. These methods could help us automatically locate relevant fixes quickly when being attacked. Our tools will in general enable better design of incident response and incident reporting requirements for organizations, by showing bottlenecks and unused capabilities in the management of vulnerabilities and attacks.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2008
- Accession Number
- ADA486776
Entities
People
- Eric Sjoberg
- Neil C. Rowe
- Paige Adams
Organizations
- Naval Postgraduate School