Sequence Labeling Parsing by Learning Across Representations

Abstract

We use parsing as sequence labeling as a common framework to learn across constituency and dependency syntactic abstractions. To do so, we cast the problem as multitask learning (MTL). First, we show that adding a parsing paradigm as an auxiliary loss consistently improves the performance on the other paradigm. Secondly, we explore an MTL sequence labeling model that parses both representations, at almost no cost in terms of performance and speed. The results across the board show that on average MTL models with auxiliary losses for constituency parsing outperform single-task ones by 1.14 F1 points, and for dependency parsing by 0.62 UAS points.

arXiv commentProc. of the 57th Annual Meeting of the Association for Computational Linguistics (ACL 2019). Revised version after fixing evaluation bug

References in corpus (4)

• An Overview of Multi-Task Learning in Deep Neural Networks
• Introduction to the CoNLL-2002 Shared Task: Language-Independent Named Entity Recognition
• Grammar as a Foreign Language
• Introduction to the CoNLL-2000 Shared Task: Chunking

Cited by in corpus (5)

• Multitask Pointer Network for Multi-Representational Parsing
• Mimic and Conquer: Heterogeneous Tree Structure Distillation for Syntactic NLP
• A Syntax-Injected Approach for Faster and More Accurate Sentiment Analysis
• Multi-Task Learning by using Contextualized Word Representations for Syntactic Parsing of a Morphologically Rich Language
• Concurrent Parsing of Constituency and Dependency