Extending a Parser to Distant Domains Using a Few Dozen Partially Annotated Examples

TL;DR

This paper demonstrates that modern word representations reduce the need for domain adaptation in parsers, and introduces a simple method to adapt parsers to distant domains using only dozens of partially annotated examples, achieving significant improvements.

Contribution

It shows that recent word representations lessen domain adaptation needs and proposes a straightforward adaptation method with minimal annotations for distant domains.

Findings

Achieved over 90% F1 on Brown corpus with a parser trained only on Wall Street Journal.

Improved geometry-domain parse accuracy from 45% to 73% with about fifty partial annotations.

Set a new state-of-the-art single model result on WSJ test set at 94.3% F1.

Abstract

We revisit domain adaptation for parsers in the neural era. First we show that recent advances in word representations greatly diminish the need for domain adaptation when the target domain is syntactically similar to the source domain. As evidence, we train a parser on the Wall Street Jour- nal alone that achieves over 90% F1 on the Brown corpus. For more syntactically dis- tant domains, we provide a simple way to adapt a parser using only dozens of partial annotations. For instance, we increase the percentage of error-free geometry-domain parses in a held-out set from 45% to 73% using approximately five dozen training examples. In the process, we demon- strate a new state-of-the-art single model result on the Wall Street Journal test set of 94.3%. This is an absolute increase of 1.7% over the previous state-of-the-art of 92.6%.