Rewriting IMDb

Coming up are some snippets and thoughts of how to convert and use the data from IMDb and make it somewhat more visual i.e. 3Dify. I’ll be using an embedded H2 database and Hibernate for persistence, Hibernate Search for searching capabilities, Ehcache to speed things up, Hadoop for con/inverting the big actor and actresses files, Mahout for clustering and suggesting data and finally Processing (PeasyCam and ControlP5) for the 3D stuff and a Wii-remote control for navigating the 3D spaced-interface.

I will be using the Wii-remote to navigate the network of actors, directors and the movies that link them all together. So the goal is to first search a movie, actor or director and then render this as a node in a 3D space after which it will start loading the connected actors, directors and movies creating some cool visuals. Surfing back and forward will be done through swinging with the controller left and right, navigating to linked nodes will be done with the up-down-left-right buttons. Free navigation through the network will be done with the Nunchuk control.

Making sense of the IMDb-data through visualization is the next step in this project leveraging the different algorithms of Mahout. This part is still being brewed in my head so stay tuned for some comments and design brainfarts about that.

Advertisements