conftrace_

2023 EMNLP EMNLP 2023

Tokengram_F, a Fast and Accurate Token-based chrF++ Derivative

Abstract

AbstractTokengram_F is an F-score-based evaluation metric for Machine Translation that is heavily in- spired by chrF++ and can act as a more accurate replacement. By replacing word n-grams with n-grams obtained from tokenization algorithms, tokengram_F better captures similarities between words.

Authors

Sören Dreano, Derek Molloy, Noel Murphy

Topics

Natural Language Processing > Applications > Information Retrieval Natural Language Processing > Applications > Machine Translation

Keywords

machine translation evaluation metric

More from EMNLP 2023

All EMNLP 2023 papers →

Findings of the WMT 2023 Shared Task on Low-Resource Indic Language Translation

ACES: Translation Accuracy Challenge Sets at WMT 2023

Challenging the State-of-the-art Machine Translation Metrics from a Linguistic Perspective

Embed_Llama: Using LLM Embeddings for the Metrics Shared Task

eBLEU: Unexpectedly Good Machine Translation Evaluation Using Simple Word Embeddings

Cometoid: Distilling Strong Reference-based Machine Translation Metrics into Even Stronger Quality Estimation Metrics