KG-BERTScore: Incorporating Knowledge Graph into BERTScore for Reference-Free Machine Translation Evaluation

TL;DR

KG-BERTScore enhances reference-free machine translation evaluation by integrating knowledge graph information with BERTScore, leading to higher correlation with human judgments than existing metrics.

Contribution

This paper introduces KG-BERTScore, a novel metric combining BERTScore with bilingual named entity matching and knowledge graph data for improved reference-free translation evaluation.

Findings

KG-BERTScore outperforms current state-of-the-art metrics in correlation with human judgments.

Incorporating knowledge graphs improves the accuracy of machine translation evaluation.

The study analyzes the impact of multilingual pre-trained models and linear combination parameters.

Abstract

BERTScore is an effective and robust automatic metric for referencebased machine translation evaluation. In this paper, we incorporate multilingual knowledge graph into BERTScore and propose a metric named KG-BERTScore, which linearly combines the results of BERTScore and bilingual named entity matching for reference-free machine translation evaluation. From the experimental results on WMT19 QE as a metric without references shared tasks, our metric KG-BERTScore gets higher overall correlation with human judgements than the current state-of-the-art metrics for reference-free machine translation evaluation.1 Moreover, the pre-trained multilingual model used by KG-BERTScore and the parameter for linear combination are also studied in this paper.