RIGA at SemEval-2016 Task 8: Impact of Smatch Extensions and Character-Level Neural Translation on AMR Parsing Accuracy
This work addresses AMR parsing accuracy for NLP researchers, offering incremental improvements through scoring extensions and a novel parser.
The paper tackled AMR parsing by introducing two smatch extensions and a character-level neural translation parser, achieving a 4% gain over the baseline with error pattern analysis and a 7% gain over word-level translation, resulting in smatch F1 scores of 62% and 67% on different test sets.
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.