Arc-Standard Spinal Parsing with Stack-LSTMs
This is an incremental improvement for computational linguistics researchers working on parsing tasks.
The paper tackles the problem of parsing spinal trees (a dependency representation of constituent trees) by developing a neural transition-based parser using Stack-LSTMs. The result shows that the model adapts to different dependency relation styles but this choice minimally affects constituent structure prediction, indicating LSTMs inherently induce useful states.
We present a neural transition-based parser for spinal trees, a dependency representation of constituent trees. The parser uses Stack-LSTMs that compose constituent nodes with dependency-based derivations. In experiments, we show that this model adapts to different styles of dependency relations, but this choice has little effect for predicting constituent structure, suggesting that LSTMs induce useful states by themselves.