aliker

README

aliker

Find similar things on tumblr

Conjecture

The things that you're likely to like are the things that are liked by the people who like the things that you like.

Algorithm

• Initialize an empty map of posts to a list of users: M
• Accept post from some blog: P
• Find every user who has left a note on that post (either a like or reblog) U0 - Un
• For each user U0-Un: U
  • For each note that U wrote to some post Q:
    • Append U to the list M[Q]
• (TODO: somehow normalize this map so that some peoples' counts mean more than others -- not yet implemented)
• Initialize a second empty map of posts Q to counts of users liking each post: C
• For each key Q in M:
  • Set C[Q] to the length of the list in M[Q]
• Now, the largest values in C have the most similar keys (posts) to P

Implementation

It's in Go. It uses websockets, bootstrap, jQuery and the gorilla web toolkit.