How to detect duplicate text with some fuzzyness

Question

Some thing ago, I write small script using Text::DeDupe to remove duplicates of blog posts before I have to lay my eyes on them.

After reading Syntactic Clustering of the Web paper on which implementation is based, I would love to have ability to find overlapping documents (e.g. snippets of blogs as opposed to full text, maybe also quotes).

Do you know of any other implementation in C, C++ or perl which I can try out before writing my own?

Answer 1

SpotSigs seems to fit my bill just right, here are some references:

• http://dbpubs.stanford.edu/pub/2008-10
• http://infoblog.stanford.edu/2008/08/spotsigs-are-stopwords-finally-good-for.html
• http://ilpubs.stanford.edu:8090/860/

The soruce code for this module is hosted on GitHub:

http://github.com/jzawodn/perl-text-spotsig