3 11 2014

Wiki clustering ideas:

You create a new page. When finished, link it with existing pages using the following algorithm:

Sort existing pages by number of in-bound links (we model the page as if it could be any existing page at this step). Each skip creates an anti-link. Each link creates a link. For each added link, recompute the most likely links (for the first link model the new page as if it could be any page linked to the first linked page).

Cluster based on links and anti-links.

Got The Family for Joining Posyt, saying I need to consider my family in these decisions. Got Great Power for working on DraftPunk at Sailthru. I think I may have reached a decision, but I will let Chris retort. But perhaps I have found my own way now.

Can totally store and restore the state of Arbor.js to do lazy clustering that we offload onto the client side. Fucking awesome.

10:57:20 AM: Dreamed Josh brought back a beautiful Irish woman who I wanted to steal from him. I betrayed a man by having him sign a trick contract and then escaped through a chimney from him and snuck out the back of the building. It reminded me of living in hightower again, I lived with friends from college but also Daria.

  • (♥) I love Ian so I will write to him explaining why I will leave.

  • (♥) I love Josh so I will write to him to explain why I’ve had so little time for taijutsu.

  • (♥) I love the support team so I will find the issue in 11720 and fix it.