Google Summer of Code

What a TRIP!

I’ve just submitted my Google Summer of Code (GSoC) application. I’m applying to the Apache Software Foundation‘s Mahout project. I’m proposing to implement a distributed recommender algorithm based on a mathematical technique called Singular Value Decomposition.

So what is GSoC? Basically Google for a number of years now Google having been helping open source projects by funding roughly 1000 summer placements. The students work with the open source organisations and code from the end of may until mid aug. The student gets $4500 US, the organisation gets $500 and some nice new shiny code, documentation or whatever else the student has been doing.

That’s the very minimum of background I can give you (check links for more info) to the story I actually want to tell…

Over the past two weeks since completing my coursework I have immersed myself in all things Mahout and all things Recommender Systems especially SVD-based recommenders. Knowing that I wanted to implement a Recommender system, and with some notes of interest from the Mahout community I decided I’d do an SVD-based algorithm. From there I trawled the literature choosing the SVD++(pdf) algorithm. After that I read, and I read and I read. Not just research papers and websites, but also code and documentation. My goal to delve into Mahout, Hadoop and recommender algorithms enough that I could write and implement my GSoC proposal.

Finally I got to a place where I understood it, at least enough to meaningfully write the proposal. With a couple of days before the application deadline I decidced to let it rest for a bit before finishing it up and submitting it. Then today on the train up to Scotland with 3 hours to go I started trying to submit it.

THREE HOURS!!! Connection drops, timeouts, an error in my application that meant it needed resubmitted and a system crash later I was out of time. AAAAARRRRGH! The rage! The frustration. Even after time I still tried on the off chance I would be allowed to submit late. About 30 minutes after the deadline the connection became crystal clear, internet was working perfectly.

As quick as I could I went to submit. As expected submissions no longer available. So, all was lost! Unless… unless if by chance one of the failed attempts had actually gone through. Holding my breath I click the ‘List my proposals link’. Slowly the page loaded… YES! WOOOOOOOHA! There was my beautiful, beautiful proposal! The relief! The excitement! I’m still in the race. and what did the submission time stamp say? 18:44. An hour and 15 before the deadline. Haha!

Here’s hoping I get it!