BERTScore: Evaluating Text Generation with BERT

Authors: Zhang et al. Year: 2019 ArXiv/Link: https://arxiv.org/abs/1904.09675

Summary

Semantic similarity metric using BERT embeddings, correlating better with human judgment than BLEU.

Key Concepts

• Semantic evaluation
• BERT embeddings
• Generation metrics
• Human correlation
• Quality assessment

Impact

Better evaluation metric for text generation tasks

Category

Evaluation Frameworks

---