Frame-Semantic Parsing with Softmax-Margin Segmental RNNs and a Syntactic Scaffold

TL;DR

This paper introduces an efficient frame-semantic parser that leverages a segmental RNN and a syntactic scaffold from Penn Treebank annotations, achieving state-of-the-art results without relying on syntactic parsing at inference.

Contribution

The paper proposes a novel syntactic scaffold training method that improves frame-semantic parsing performance without requiring syntactic parsing during testing.

Findings

Achieves competitive performance without syntactic parser calls.

Uses Penn Treebank annotations as a training scaffold.

Attains state-of-the-art results in frame-semantic parsing.

Abstract

We present a new, efficient frame-semantic parser that labels semantic arguments to FrameNet predicates. Built using an extension to the segmental RNN that emphasizes recall, our basic system achieves competitive performance without any calls to a syntactic parser. We then introduce a method that uses phrase-syntactic annotations from the Penn Treebank during training only, through a multitask objective; no parsing is required at training or test time. This "syntactic scaffold" offers a cheaper alternative to traditional syntactic pipelining, and achieves state-of-the-art performance.