The reviewed record of science sign in
Pith

arxiv: 2108.05682 · v2 · pith:OFV2HG6M · submitted 2021-08-12 · cs.CL

(Un)solving Morphological Inflection: Lemma Overlap Artificially Inflates Models' Performance

Reviewed by Pithpith:OFV2HG6Mopen to challenge →

classification cs.CL
keywords modelsinflectionlanguagesaveragepointschallengedropeven
0
0 comments X
read the original abstract

In the domain of Morphology, Inflection is a fundamental and important task that gained a lot of traction in recent years, mostly via SIGMORPHON's shared-tasks. With average accuracy above 0.9 over the scores of all languages, the task is considered mostly solved using relatively generic neural seq2seq models, even with little data provided. In this work, we propose to re-evaluate morphological inflection models by employing harder train-test splits that will challenge the generalization capacity of the models. In particular, as opposed to the na{\"i}ve split-by-form, we propose a split-by-lemma method to challenge the performance on existing benchmarks. Our experiments with the three top-ranked systems on the SIGMORPHON's 2020 shared-task show that the lemma-split presents an average drop of 30 percentage points in macro-average for the 90 languages included. The effect is most significant for low-resourced languages with a drop as high as 95 points, but even high-resourced languages lose about 10 points on average. Our results clearly show that generalizing inflection to unseen lemmas is far from being solved, presenting a simple yet effective means to promote more sophisticated models.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.