Span-Based Constituency Parsing with a Structure-Label System and Provably Optimal Dynamic Oracles
This work addresses the challenge of making neural models competitive in constituency parsing, which is incremental as it builds on existing transition systems but introduces key innovations for efficiency and accuracy.
The paper tackles the problem of improving constituency parsing accuracy with neural models by introducing a new span-based shift-reduce system and a provably optimal dynamic oracle, achieving the best F1 scores on English and French without reranking or external data.
Parsing accuracy using efficient greedy transition systems has improved dramatically in recent years thanks to neural networks. Despite striking results in dependency parsing, however, neural models have not surpassed state-of-the-art approaches in constituency parsing. To remedy this, we introduce a new shift-reduce system whose stack contains merely sentence spans, represented by a bare minimum of LSTM features. We also design the first provably optimal dynamic oracle for constituency parsing, which runs in amortized O(1) time, compared to O(n^3) oracles for standard dependency parsing. Training with this oracle, we achieve the best F1 scores on both English and French of any parser that does not use reranking or external data.