Completed/Retired - FidoFetch is a news reader that shows you articles interesting to you by learning about what you like.

##What is it?## FidoFetch is a RSS/ATOM news reader service much like Google Reader. There are too many news articles and blog posts every second to read it all. In fact, most of it is not something you care about. FidoFetch watches which kinds of articles you like and only shows you things it thinks you will be interested in.

##When## Over the past 5+ years I have been working on this idea in my spare time. I’ve been researching approaches to the problem and have implemented several different versions to test out ideas. I am currently working on a system to test several different machine learning algorithms at the same time to evaluate which works the best.


  • The first implementation was done in PHP with MySQL as the backend.
  • The first implementation used Bayes’ Theorem for filtering the news.
  • The third implementation uses PHP Python for the front-end and Cassandra Riak as the backend.
  • I have used Hadoop to do some analytics on data I collected.

If you want a more in-depth look at how FidoFetch works, checkout my detailed post on FidoFetch’s Architecture.

##Success?## The first implementation was able to figure out the taste of my friends and myself however it was confused with common words. I am working on other algorithms that take into account common words.

Check out FidoFetch and let me know what you think!