11 citations · 16 across the 6 of their papers we have counts for
1 paper · 2 filters
Philipp Meerkamp, Zhengyi Zhou
We present an architecture for information extraction from text that augments an existing parser with a character-level neural network. The network is trained using a measure of co…