Greedy, Joint Syntactic-Semantic Parsing with Stack LSTMs

TL;DR

This paper introduces a greedy, transition-based parser that jointly models syntactic and semantic dependencies using stack LSTMs, achieving state-of-the-art results efficiently on English shared tasks.

Contribution

It presents a novel joint parsing model with a learned representation of parser state using stack LSTMs, enabling linear-time inference.

Findings

Achieved best published joint syntax-semantics parsing performance on CoNLL 2008-9.

Developed a linear-time greedy inference algorithm.

Utilized stack LSTMs for comprehensive parser state representation.

Abstract

We present a transition-based parser that jointly produces syntactic and semantic dependencies. It learns a representation of the entire algorithm state, using stack long short-term memories. Our greedy inference algorithm has linear time, including feature extraction. On the CoNLL 2008--9 English shared tasks, we obtain the best published parsing performance among models that jointly learn syntax and semantics.