Tetra-Tagging: Word-Synchronous Parsing with Linear-Time Inference

Nikita Kitaev, Dan Klein
4/22/2019
cs.CL

Official code: nikitakit/tetra-tagging (14 stars, last commit 2020-07-06).

Abstract

We present a constituency parsing algorithm that, like a supertagger, works by assigning labels to each word in a sentence. In order to maximally leverage current neural architectures, the model scores each word's tags in parallel, with minimal task-specific structure. After scoring, a left-to-right reconciliation phase extracts a tree in (empirically) linear time. Our parser achieves 95.4 F1 on the WSJ test set while also achieving substantial speedups compared to current state-of-the-art parsers with comparable accuracies.

View on arXivView PDF

Code Implementations(1)

nikitakit/tetra-taggingOfficial100%

Tetra-Tagging: Word-Synchronous Parsing with Linear-Time Inference

142PythonMay 2, 20206 years ago

Similar Papers

Revisiting Supertagging for Faster HPSG Pasing

84% match2023

Constituency Parsing with a Self-Attentive Encoder

83% match1 repo9132018

Cross-Lingual Constituency Parsing for Middle High German: A Delexicalized Approach

81% match2023

Representing and Parsing Korean Constituency Structure at Different Levels of Granularity

81% match2026

TreePiece: Faster Semantic Parsing via Tree Tokenization

80% match2023

Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Autoencoders

80% match1 repo892019

Cite this paper

@article{kitaev2019tetratagging,
  title  = {Tetra-Tagging: Word-Synchronous Parsing with Linear-Time Inference},
  author = {Nikita Kitaev and Dan Klein},
  year   = {2019},
  eprint = {1904.09745},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1904.09745}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1904.09745.svg)](https://codeofpaper.com/paper/1904.09745)

Discussion