Tech Report CS-09-01
Modeling Valence Effects in Unsupervised Grammar Induction
We extend the dependency grammar induction model of Klein and Manning (2004) to incorporate further valence information. Our extensions achieve significant improvements in the task of unsupervised dependency grammar induction. We use an expanded grammar which tracks higher orders of valence and allows each valence slot to be filled by a separate distribution rather than using one distribution for all slots. Additionally, we show that our performance improves if our grammar restricts the maximum number of attachments in each direction, forcing our system to focus on the common case. Taken together, these techniques constitute a 23.4% error reduction in dependency grammar induction over the model by Klein and Manning (2004) on English.
(complete text in pdf)