Acronym Identification: 10K-100K Expert-Annotated Scientific Sentences

Between 10,000 and 100,000 expert-annotated sentences comprise this dataset for token-level acronym identification in the scientific domain. Created by Amirveyseh for the AAAI-21 Workshop on Scientific Document Understanding, it includes standardized training, validation, and test splits.

Use Cases

• Training token classification models to identify acronym labels within scientific text
• Benchmarking sequence labeling algorithms using the provided training, validation, and test splits
• Developing information extraction systems to isolate acronym tokens from pre-tokenized sentences

Strengths

• Expert-generated annotations from the AAAI-21 SDU workshop
• Scale of 10,000 to 100,000 records
• MIT license allows for open research and commercial use

Limitations

• Restricted to scientific document domain
• Monolingual English coverage

Provenance

Source

AAAI-21 Workshop on Scientific Document Understanding

Collection Method

Expert-annotated

Time Range

2021

Freshness

Last updated January 2024