We just finished a little thing for the Universal Music Group. You might call it a "social commerce" app if you're that way inclined.

Here's how it works:

• User connects through Facebook (graph oAuth)
• We scrape their music 'likes'
• We compare those to the sites catalogue (made up of products from the Universal catalogue)
• We present recommendations based on the user's likes

You can also pick a friend which follows a similar path.

So basically, we're presenting product choices based on likes. This allows us to present options that are, in theory, nicely targetted. This is even nicer when you pick a friend through the app. Ever struggled to decide what to buy your mum or brother for Christmas? If they've 'liked' music, either through Facebook or from an off-Facebook Like Button we have a good chance of recommending something that they'll be into.

That was actually the easy bit. The difficult bit was automatically broadening the 'like vs. catalogue' match possibilities. For example, my Facebook profile might list me as being interested in Arcade Fire, but what if the Universal catalogue available to the site doesn't include any Arcade Fire albums?

To broaden the match possibilities, we decided to use the Amazon Catalogue API. It's really great. We pass in an ASIN (unique Amazon product ID) and ask it for similar artists. Because Amazon's recommendation engine is the best in the world, it gives us lots of really accurate artist similarities. We store the similar artists alongside the products from Universal's catalogue, so when Facebook tells the app that I like Arcade Fire, there's a very good chance we'll find some product matches.

The catalogue's still being built up, and the bigger the catalogue, the better the results, but it's looking good so far.

To hit the Christmas shopping period, we built the app in just under two weeks (10 working days).

givemusic.co.uk