Large-Scale QA-SRL Parsing

Nicholas FitzGerald, Julian Michael, Luheng He, Luke Zettlemoyer
5/14/2018
cs.CLcs.AI

Official code: uwnlp/qasrl-bank (21 stars, MIT, last commit 2021-02-13).

Abstract

We present a new large-scale corpus of Question-Answer driven Semantic Role Labeling (QA-SRL) annotations, and the first high-quality QA-SRL parser. Our corpus, QA-SRL Bank 2.0, consists of over 250,000 question-answer pairs for over 64,000 sentences across 3 domains and was gathered with a new crowd-sourcing scheme that we show has high precision and good recall at modest cost. We also present neural models for two QA-SRL subtasks: detecting argument spans for a predicate and generating questions to label the semantic relationship. The best models achieve question accuracy of 82.6% and span-level accuracy of 77.6% (under human evaluation) on the full pipelined QA-SRL prediction task. They can also, as we show, be used to gather additional annotations at low cost.

View on arXivView PDF

Code Implementations(1)

uwnlp/qasrl-bankOfficial100%
MIT

Central repository for QA-SRL data.

213ShellMay 2, 20185 years agoMIT

Similar Papers

Schema2QA: High-Quality and Low-Cost Q&A Agents for the Structured Web

83% match3 repos2022020

SearchQA: A New Q&A Dataset Augmented with Context from a Search Engine

83% match1 repo1812017

emrQA: A Large Corpus for Question Answering on Electronic Medical Records

83% match1 repo1532018

ELI5: Long Form Question Answering

81% match1 repo3262019

KGConv, a Conversational Corpus grounded in Wikidata

80% match2023

Addressing Semantic Drift in Question Generation for Semi-Supervised Question Answering

80% match1 repo972019

Cite this paper

@article{fitzgerald2018largescale,
  title  = {Large-Scale QA-SRL Parsing},
  author = {Nicholas FitzGerald and Julian Michael and Luheng He and Luke Zettlemoyer},
  year   = {2018},
  eprint = {1805.05377},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1805.05377}
}

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/1805.05377.svg)](https://codeofpaper.com/paper/1805.05377)

Discussion