The next lab meeting will be on Monday October 17, 2011 at 3:30pm, in TASC1 9408.
Marzieh Razavi will be giving a talk which she describes as follows:
I’ll be talking about two clustering methods (Brown Clustering and HMM State Splitting) to build a discriminative dependency parser in the MSTParser format using cluster-based features. The clustering-based parsers are combined together using the Ensemble model. Diverse clustering-based parser models and the ensemble method together significantly improves unlabeled dependency accuracy from 91.07% to 92.37% on Section 23 of the Penn Treebank.