SemBleu: A Robust Metric for AMR Parsing Evaluation

TL;DR

SEMBLEU is a new AMR parsing evaluation metric extending BLEU, which avoids search errors and better captures non-local correspondences, showing higher consistency with human judgments than SMATCH.

Contribution

It introduces SEMBLEU, a robust, content-driven metric for AMR parsing evaluation that improves upon SMATCH by eliminating search errors and considering non-local matches.

Findings

SEMBLEU has higher consistency with human judgments than SMATCH.

SEMBLEU is fully content-driven and avoids search errors.

Preliminary experiments show SEMBLEU performs well at sentence and corpus levels.

Abstract

Evaluating AMR parsing accuracy involves comparing pairs of AMR graphs. The major evaluation metric, SMATCH (Cai and Knight, 2013), searches for one-to-one mappings between the nodes of two AMRs with a greedy hill-climbing algorithm, which leads to search errors. We propose SEMBLEU, a robust metric that extends BLEU (Papineni et al., 2002) to AMRs. It does not suffer from search errors and considers non-local correspondences in addition to local ones. SEMBLEU is fully content-driven and punishes situations where a system's output does not preserve most information from the input. Preliminary experiments on both sentence and corpus levels show that SEMBLEU has slightly higher consistency with human judgments than SMATCH. Our code is available at http://github.com/freesunshine0316/sembleu.