Stylometry… stylometry
is the way in which a person writes; and that style allows it to be quantified
and identified. It is the way in which a person creates a paragraph, the length
of sentences, the use of vocabulary… all this contributes to obtain a
scientific perspective of the style of writing of any single person using an
electronic device. In OPSEC terms (Operations Security), stylometrics are a
subject that deserves full attention because it is very difficult to defend and
protect ourselves from them.
One of the
operational safety measures that Edward Snowden applied involved writing small
paragraphs and short sentences, but never full and well-structured texts. The
reason for this measure was that regardless of the types of anonymity systems
being used, if any kind of spyware or governmental agency were able to extract
a plaintext, this action could expose him and relate it to some other documents
previously written. The NSA´s repertoire is massive, but Semantic Analysis is by far the scariest.
Any good story
must have dreamers and geniuses, villains and visionaries… the first company
that came out with this type of software was not the exception to this rule and
obtained the whole and undivided attention from governmental agencies, mainly
the CIA.
Palantiri (One that Sees from Afar) is “a fictitious” artifact of the legendarium created by Daddy J.R.R. Tolkien and is a spherical stone that serves to see distant events or places or to communicate with the user of another stone.
"Reality is stranger than fiction..."
“They're in a scary business”. –Lee Tien, EFF.
Palantir, which is very real, was founded back
in 2004 by Peter Thiel (PayPal), Alex Karp, Joe Lonsdale, Stephen Cohen and
Nathan Gettings… with an initial investment of $2 million USD through one of
the CIA´s main companies, In-q-Tel.
But it does not stop there because its main clients are
the NSA, the CIA, the FBI, NewsCorp and JPMorgan Chase. Their advisers are
Condoleezza Rice, George Tenet & General David Petraeus.
"Any war against humanity is betrayal"
¨I do not expect to see home again¨. –Edward Snowden
In five years Palantir became the company to which “The Dark One” goes when in need of “Massive Data Mining”, and unlike other agencies and companies Palantir realized that the single use of Artificial intelligence would work against an “adaptable adversary”, that´s when they started using teams of human analysts to explore data being originated from different sources, this process is called intelligence augmentation.
In 2007 The University of Arizona sneaked into blogs, forums and any
platform in the Deep Web that allowed them to gather information, the idea was
to compile and analyze all that delicious data… obviously using stylometrics.
"How?”
There are two types of analysis: Supervised and Unsupervised. In Supervised Analysis, there is a text with a totally identifiable author and different "anonymous texts", they compare them and look for matching fingerprints (writing style). For Unsupervised Analysis, the computer reviews complete blocks of anonymous texts and it piles them up according to different parameters in writing styles: A, B, C…
There are two types of analysis: Supervised and Unsupervised. In Supervised Analysis, there is a text with a totally identifiable author and different "anonymous texts", they compare them and look for matching fingerprints (writing style). For Unsupervised Analysis, the computer reviews complete blocks of anonymous texts and it piles them up according to different parameters in writing styles: A, B, C…
The document
presented by Michael Brennan, Sadia Afroz & Rachel Greenstadt at 28C3 shows 2 ways to avoid “being
labeled”:
1. Obfuscation: To make a text resemble something that is definitely not
your style… for example, translate the text to one language and back to the
language of origin using google translate.
2.
Imitation: Make your style
resemble someone else´s… Donald Trump, you are so screwed!
Anonymouth is
simply beautiful… once you have mastered it. Anonymouth is a Java-based tool, so always keep in mind that requires
constant updating. The way it works is by comparing the desired text to other
documents written by the same author; the outcome is a neat breakdown of used
common words among all texts, sentence structure and proper suggestions for
modifications.
Stylometrics are this century's version of phrenology
PRISM is a government
code name for a data-collection effort known officially by the SIGAD US-984XN, PRISM allows the NSA to connect itself directly to the central servers of
“The Big 9”: Microsoft, Yahoo, Google, Facebook, PalTalk, AOL, Skype, YouTube,
Apple… in the meanwhile, Palantir keeps growing.
Published on 10/10/14 by @feexitmx
Translated by DM
Translated by DM
No comments:
Post a Comment