RIGA at SemEval-2016 Task 8: Impact of Smatch Extensions and Character-Level Neural Translation on AMR Parsing Accuracy

TL;DR

This paper introduces extensions to the AMR smatch scoring method, improving parsing accuracy by error analysis, ensemble selection, and character-level neural translation, achieving significant F1 score gains.

Contribution

It presents novel smatch extensions including error pattern analysis and ensemble methods, and demonstrates the effectiveness of character-level neural translation for AMR parsing.

Findings

4% improvement with error pattern analysis and error correction.

0.4% gain using ensemble selection among multiple AMR graphs.

7% increase in accuracy with character-level neural translation.

Abstract

Two extensions to the AMR smatch scoring script are presented. The first extension com-bines the smatch scoring script with the C6.0 rule-based classifier to produce a human-readable report on the error patterns frequency observed in the scored AMR graphs. This first extension results in 4% gain over the state-of-art CAMR baseline parser by adding to it a manually crafted wrapper fixing the identified CAMR parser errors. The second extension combines a per-sentence smatch with an en-semble method for selecting the best AMR graph among the set of AMR graphs for the same sentence. This second modification au-tomatically yields further 0.4% gain when ap-plied to outputs of two nondeterministic AMR parsers: a CAMR+wrapper parser and a novel character-level neural translation AMR parser. For AMR parsing task the character-level neural translation attains surprising 7% gain over the carefully optimized word-level neural translation. Overall, we achieve smatch F1=62% on the SemEval-2016 official scor-ing set and F1=67% on the LDC2015E86 test set.