Book Review: Programming Collective Intelligence

Having never written a book review before (well maybe at school, but nothing in any recent memory) I'm not really sure to start with this - so maybe I'll go completely arse-about-face and decide to post my verdict first...

Verdict

If you have any interest in how things like Google or spam-filtering works, BUY THIS BOOK. The descriptions are clear and there are working code samples to help clarify the descriptions.

Explanation

The book contains many chapters on things such as how systems such as Amazon recommend products based on purchases by other users and how search engines work. The cool thing is though that each section goes beyond the basics but keeps the descriptions and (Python) code clean enough to understand.

For example, considering the two examples above the recommendation chapter describes not just matching up "you bought X, user B also bought X and he also bought Y so therefore you might like Y" but mapping how close your tastes match on other books therefore how likely the recommentation will be successful. Likewise, the search chapter takes you from basic tokenising of content, through determining how close in distance multiple terms appear through to a simple implementation of Google's PageRank algorithm.

As a note of warning, the book isn't light reading (I first started reading it at about 10:30pm and decided it would be better to start in the morning with a clear head) but it's not as heavy as other text books on the topics go (I've recently got in to modern information retrieval systems as part of my current contract with IPC Media).

Buy the book at Amazon.co.uk.