Week 10

I've written up a report on my work and the results, which can be found here.

My work isn't quite done, though. We're going to look into figuring out which of the specific parse features were the most useful, and look into how to tell LinkGrammar to only parse one linkage (since right now it's doing as many as it can find, even though we're only using one. This should help improve run-time.), and also try running the code with an order of two, which is the more common way for these conditional random fields sorts of NER systems to go. After that additional work, I'll write up a longer paper to be submitted to a conference.

It's been a lot of fun, and an amazing experience. I was really intimidated by the prospect of doing research before, and now I feel like I've got a good grasp on the workings of it and graduate school. I look forward to doing more research in the future.