Link to Home
Silence Nogood title
Background Glow for Title
Background Glow for Social Networking Links

The Hype Machine Should Go Green

And when I say green, I mean organic... in which I really mean automated.

Step 1: Mining for Recommendation

The Hype Machine allows you to manually subscribe to the artists, blogs, friends and search terms of your choice. As helpful as it’s been to create a more customized playlist, it becomes increasingly harder to keep up with and subscribe to all the stuff you like. Therefore, Hypem should start building an automated recommendation system to do the task for you. Before we can figure out what to recommend and how to display it on Hypem – which’ll be covered in a later post – we’ll need to find out where to pull the information from. Luckily, it’s right at the heart of The Hype Machine… literally, the Hypem Heart

When you ‘heart’ a song on Hypem, you’re saying ya like it, ya love it, it’s your favorite, whatever – but that’s not all. Certain characteristics of the song can also be extracted to help build a recommendation system- these characteristics are its tags, the blogs that posted it & other users who’ve ‘hearted’ it.

Hypem loves

Tags are one of the most accurate, flexible ways to mark similarities between songs. However, the accuracy comes down to the size & enthusiasm of the community of taggers. Since Hypem pulls its tags from the community, it’s fair to assume it’s about as good as it’s gonna get for now. But as helpful as these tags are for matching simple similarities, it starts to get a little more interesting when we incorporate Hypem’s community into the equation.

Incorporating the blogs that posted the song and other users who’ve ‘hearted’ it, will make the recommendation system all the more unique to Hypem, finding new and never before made connections. It’s important to note that these types of characteristics need to be carefully weighted – i.e. just because you liked an Emancipator song posted by a blog, doesn’t mean you’ll like their next post. But with the appropriate testing, it could make Hypem’s recommendation system one of a kind.

Why hasn’t it been implemented and where we’re going from here

Now, the idea of an automated recommendation system isn’t all that original. The reason it isn’t up on Hypem yet is probably due to the many implementation issues that arise (e.g. coding, costs, etc.). The reason I decided to bring it up is because (a.) I wanted to start looking into how music services work & how to improve on them (with my newly created UX Design category) and (b.), more importantly, this sets up my next article on Hypem, dealing with the front page and ways to revamp it.

So let us know what you’d think if Hypem implemented a “recommendation system” and any ideas you have on improving its front page (maybe I’ll include ’em in the next article!)

Hype Machine Stickers