What question did this study set out to answer?

To develop and assess an efficient named entity recognition model for legal texts that operates effectively with minimal annotated training data.

June 19, 2023

LeArNER

View Full Paper

Ask AI

Bookmark

Authors

Computational evaluation demonstrates a 94.88% F1 score in Traditional Chinese legal texts, indicating high efficiency with minimal annotated data.

To develop and assess an efficient named entity recognition model for legal texts that operates effectively with minimal annotated training data.
Trained the LeArNER framework using an optimized sample size of 2,000 sentences.
Evaluated performance on a corpus of Traditional Chinese constitutional legal cases from Taiwan across 13 target entity types.
Achieved an overall F1 score of 94.88% across all 13 legal entity categories.
Peak performance was attained using a training set of only 2,000 sentences, demonstrating low data dependency.

Lee et al. (2023) studied this question.

View Full Paper

Ask AI

Bookmark

Computational evaluation demonstrates a 94.88% F1 score in Traditional Chinese legal texts, indicating high efficiency with minimal annotated data.