Unsupervised Dependency Parsing: Let's Use Supervised Parsers
This work addresses the problem of improving parsing accuracy in natural language processing for researchers, but it is incremental as it builds on existing supervised and unsupervised methods.
The paper tackles unsupervised dependency parsing by proposing a self-training approach called iterated reranking, which improves trees from an unsupervised parser using supervised models trained on those trees, achieving a 1.8% higher accuracy than the state-of-the-art on the WSJ corpus.
We present a self-training approach to unsupervised dependency parsing that reuses existing supervised and unsupervised parsing algorithms. Our approach, called `iterated reranking' (IR), starts with dependency trees generated by an unsupervised parser, and iteratively improves these trees using the richer probability models used in supervised parsing that are in turn trained on these trees. Our system achieves 1.8% accuracy higher than the state-of-the-part parser of Spitkovsky et al. (2013) on the WSJ corpus.