5 ms·

Automatic transformation of text to evade these methods seems feasible (google translate back and forth might be the crude first attempt.) Obviously there might

by praptak 13y ago

Automatic transformation of text to evade these methods seems feasible (google translate back and forth might be the crude first attempt.) Obviously there might exist more refined methods of identification. In case of a book it is probably hard not to ruin it this way but reviews, posts and such do not require such high standards.

ledge 13y ago
There are a few open source "adversarial stylometry" software projects, check out JStylo/Anonymouth.
coldarchon 13y ago
Won't help. I wrote a bot to understand NLP and tought him texts from Shakespeare and a lot other stuff in the free domain. He could tell he was given a sentence by Shakespeare just by the structure of the sentence. I used a package called Moby. It must be much easier if you pick a whole book.
gwern 13y ago
http://www.cs.drexel.edu/~sa499/papers/adversarial_stylometry.pdf http://www.cs.drexel.edu/~sa499/papers/adversarial_stylometr... "Adversarial Stylometry: Circumventing Authorship Recognition to Preserve Privacy and Anonymity" Brennan et al 2013 Machine translation doesn't work at all. Imitation or deliberate changes to style does - but if Rowling were to do that, she would probably sacrifice a lot of quality or effort and it would defeat the apparent point of the experiment (to demonstrate that she wrote good books, as assessed by 'blinded' reviewers).
eshvk 13y ago
It is very easy to ruin a book even with manual translation. E.g. Albert Camus's The Stranger was first translated by some old English guy who brought in significant cultural aspects to the book that didn't exist in the french. Google translate is nice enough for the quick grokking of post but it is nowhere close to even the badness of manual.
eru 13y ago
Note to coldarchon: It seems like you are hellbanned..
elchief 13y ago
Here's a video series I made a couple of years ago on text mining with RapidMiner: http://vancouverdata.blogspot.ca/2010/11/text-analytics-with-rapidminer-loading.html http://vancouverdata.blogspot.ca/2010/11/text-analytics-with... It's pretty easy to compare document similarity with n-grams of various lengths. Did some work with a law firm to do something similar.

by praptak 13y ago

ledge 13y ago

There are a few open source "adversarial stylometry" software projects, check out JStylo/Anonymouth.

coldarchon 13y ago

Won't help. I wrote a bot to understand NLP and tought him texts from Shakespeare and a lot other stuff in the free domain. He could tell he was given a sentence by Shakespeare just by the structure of the sentence. I used a package called Moby. It must be much easier if you pick a whole book.

gwern 13y ago

http://www.cs.drexel.edu/~sa499/papers/adversarial_stylometry.pdf http://www.cs.drexel.edu/~sa499/papers/adversarial_stylometr... "Adversarial Stylometry: Circumventing Authorship Recognition to Preserve Privacy and Anonymity" Brennan et al 2013 Machine translation doesn't work at all. Imitation or deliberate changes to style does - but if Rowling were to do that, she would probably sacrifice a lot of quality or effort and it would defeat the apparent point of the experiment (to demonstrate that she wrote good books, as assessed by 'blinded' reviewers).

eshvk 13y ago

It is very easy to ruin a book even with manual translation. E.g. Albert Camus's The Stranger was first translated by some old English guy who brought in significant cultural aspects to the book that didn't exist in the french. Google translate is nice enough for the quick grokking of post but it is nowhere close to even the badness of manual.

eru 13y ago

Note to coldarchon: It seems like you are hellbanned..

elchief 13y ago

Here's a video series I made a couple of years ago on text mining with RapidMiner: http://vancouverdata.blogspot.ca/2010/11/text-analytics-with-rapidminer-loading.html http://vancouverdata.blogspot.ca/2010/11/text-analytics-with... It's pretty easy to compare document similarity with n-grams of various lengths. Did some work with a law firm to do something similar.