Incremental Parsing with Minimal Features Using Bi-Directional LSTM
This work addresses the need for minimal feature engineering in parsing tasks for computational linguistics, though it appears incremental as it builds on existing neural network approaches.
The authors tackled the problem of reducing feature engineering in parsing by using bi-directional LSTM sentence representations with only three sentence positions, achieving state-of-the-art results among greedy dependency parsers for English and also introducing a novel transition system for constituency parsing that achieves state-of-the-art results for English and Chinese.
Recently, neural network approaches for parsing have largely automated the combination of individual features, but still rely on (often a larger number of) atomic features created from human linguistic intuition, and potentially omitting important global context. To further reduce feature engineering to the bare minimum, we use bi-directional LSTM sentence representations to model a parser state with only three sentence positions, which automatically identifies important aspects of the entire sentence. This model achieves state-of-the-art results among greedy dependency parsers for English. We also introduce a novel transition system for constituency parsing which does not require binarization, and together with the above architecture, achieves state-of-the-art results among greedy parsers for both English and Chinese.