Rico is a DJ and is lazy and would rather code a bot to search for new artists for him. Sean thought the anime waifu pillow was silly and joined the projects. We like anime remixes so this tool helps us find similar artists, especially where the Japanese language barrier may make it tricky to..

What it does

It uses an recursive algorithm based weighting mutual following connections between artists on SoundCloud to find related artists "filling in the gaps" between the given "base" artists. Mutual connections are used to generate 'relevancy' weights which are summed up per-connection to derive the list of mutually followed artists who are the "most relevant" to the input artists. The most relevant artists are added to the list and the algorithm reiterates. The output is a beautiful, mobile, social, and local friendly revolutionary web based output which gives the user a list of artists most like the ones they inputted.

How we built it

We started off completely separately developing the algorithm that scanned SoundCloud for the connections as the website until about halfway into the Hackathon. Then we started to get the two sides of the project to intermesh and now we have a working platform.

Challenges we ran into

Understanding MVC from two completely independently developed starting points meant to merge together meant some strange compatibility issues that needed to be resolved. General inexperience with using Python for web-apps also caused some difficulties.

Accomplishments that we're proud of

An absolutely robust algorithm that has impressed musicians we know for its incredible relevancy, and the fact a functional web-app was created.

What we learned

Lots of MVC/webapp related things.

What's next for FogFlock

Tracking relevancy over time, huge data sets f, and algorithm optimization to handle several more iterations.

Share this project: