If you work for Gilt Tech than you WILL (sooner or later ... somehow) get involved in our Machine Learning and Recommendation efforts. Our requirements are extremely complex and challenging. We have over 5 million members (fast growing) and a product catalog which features products that are selling out between 10 secs and 3 days, means we need to make recommendations for products that have (almost) no history to members that might not be with us for a long time (yet). In that context we are constantly investing in making our personalized EMails and also the content of the WebSite more relevant to the member who is looking at it.

If you do/get this right, it is Mass-Customization (the art of cost-effectivly build instances of one (store)) at its best ...

“If I have 3 million customers on the Web, I should have 3 million stores on the Web.” – Jeff Bezos, CEO of Amazon.com

The difference between Amazon and Gilt is that Amazon got a "lot" of time to build these stores (hours, days, ...) . In our case we almost need to build these stores on the fly (seconds, minutes, ...). This requires new thinking, new approaches, new algorithms, ... it requires innovation. Got ideas? Let's talk!

Anyway ...

In that context I started to play with Apache Mahout and found this good (old) tutorial (still top of the list when you Google "Apache Mahout Tutorial") from Grant and decided to "maven-ize" it. Just install git and run ...

> git clone git://github.com/rolandtritsch/Apache-Mahout-Tutorial.git

... and then install maven and follow the instructions in the README. Enjoy!