In October 2010 I gave a couple of informal presentations to my lab group.

At the time I was working on some simple baseline models for supertagging. See ParsingModels for more details of this line of my research. I came up with some really simple models, which I presented in one talk. In the second talk I report on the results I got when I tried implementing them.

These baselines are all bayesian unigram models, each modelling a slightly different type of observation based on the input chords.