A Sydney security researcher has created a Twitter feed system to enrich security vulnerability news and cut through garbage tweets.
The web-based system dubbed Talkback was pitched as an alternative and more pure security news source to Twitter that skipped “utter rubbish” tweets.
It aimed to provide researchers, infosec professionals, and intrepid journalists and bloggers with easier access to rich technical documentation and commentary about vulnerabilities.
A map of vulnerability tweets
|
“Public vulnerability reports suck,” Talkback designer Matt Jones said.
He said public intelligence feeds like NVD (National Vulnerability Database) often missed important technical information because maintaining them was “an extremely hard task”.
“The real nitty-gritty of vulnerabilities is specific and requires an in-depth understanding and that can be time-consuming.”
Talkback worked by tapping tweets issued from thousands of Twitter profiles – 13,500 and growing at the time of writing – which cited vulnerability and vendor identity prefixes such as CVE (Common Vulnerabilities and Exposures), NVD, VMSA (VMware Security Advisory), and RHSA (Red Hat Security Advisory).
“They are relatively unique short character sequences so can be used to mine through the millions of messages flying through Twitter in any minute,” Jones said.
“Not only is it common for people to refer to a vulnerability by its ID, but it allows the mining process to work over many languages since these IDs are universal.”
Talkback could not only identify the first account to tweet a vulnerability reference but also those who may later tweet links to valuable blog information that shed light on vulnerabilities.
It uses social media to maintain a dynamic vulnerability timeline which is composed of a mix of “sysadmins writing about patching systems, technical write-ups, short snippets of information, and incidents and outbreaks".
But it also helps measure "the general hype from the community across all walks of life", Jones said.
Through logic algorithms and manual tweet moderation, Talkback is “learning” the difference between infosec noise and news, and is continually building its database of security researchers which it monitors.
“If you're tweeting about what you ate for lunch, then that's noise,” Jones said. “There's more quiet security people who don't tweet often, but tweet valuable information.”
Jones pointed out that knowing who to follow on social media sites like Twitter was difficult: good researchers may tweet a ream of general banter, then drop links to useful security blogs and news.
Those quiet accounts may also have few followers and be missed in the flurry of re-tweeted security news, AntiSec proclamations and daily banality.
Talkback is being modified to allow it to tap into social media sites used in countries where Twitter was not dominant, such as China.
Jones is also working on analysis of users to identify how information passed through Talkback, visualising language or location-specific vulnerability trends, and identifying where “points of influence” reside.
Jones, director of security consultancy Volvent Security and co-organiser of the Ruxcon conference, pieced Talkback together adding new features, GUIs and tweaking its smarts over the past 18 months.
The system could establish trends and help identify “hype” – popular vulnerabilities like MS12-120 and MS11-083 – which do not result in weaponised exploits or widespread compromise.
Each item is weighted by popularity from fly to obese according to the number of tweets on a given subject. The weights eliminate pollution by re-tweets, hashtags and shortened URLs by calculating the difference between parent tweets and potential child items.
A parent tweet is weighted according to the number of child items that cite it. The system determines whether child items are more or less unique.
Tweets can be represented in a side-scrolling timeline or by dots placed on a map that indicatethe location of Twitter accounts
The web interface is built with Bootstrap, jqPlot and Exhibit Simile.
Jones said in recent years more researchers have released technical documentation on bugs which appear to be driven by “independents cashing in on bug-bounties” and “research teams showcasing their technical calibre and writing about a vulnerability in-depth”.
Read Jones' blog: A year of mining Twitter.
