Learning Syntax from Naturally-Occurring Bracketings

By Tianze Shi, Ozan Irsoy, Igor Malioutov, Lillian Lee. 2021

6 citations recorded by the source corpus.

Naturally-occurring bracketings, such as answer fragments to natural language questions and hyperlinks on webpages, can reflect human syntactic intuition regarding phrasal boundaries. Their availability and approximate correspondence to syntax make them appealing as distant information sources to incorporate into unsupervised constituency parsing. But they are noisy and incomplete; to address this challenge, we develop a partial-brackets-aware structured ramp loss in learning. Experiments demonstrate that our distantly-supervised models trained on naturally-occurring bracketing data are more accurate in inducing syntactic structures than competing unsupervised systems. On the English WSJ corpus, our models achieve an unlabeled F1 score of 68.9 for constituency parsing.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jiajun Wu
2. Josh Tenenbaum
3. Jiayuan Mao
4. Roger Levy
5. Haoyue Shi
6. Christopher D. Manning
7. Tomer D. Ullman
8. Yoshua Bengio

In Natural Language Processing

1. Jiayuan Mao
2. Roger Levy
3. Josh Tenenbaum
4. Haoyue Shi
5. Jiajun Wu

Citations in the corpus

• Grammar-Based Grounded Lexicon Learning (2022, methodology): “However, our model is fundamentally different from works on unsupervised constituency parsing [23, 34] which use the CKY algorithm for inference over scalar span scores and those compute span representation vectors with CKY-style algorithms [27, 11, inter alia].”