Faceted Summaries for Long Scientific Documents

FacetSum contains over 60,000 summaries for long scientific documents, structured for faceted summarization research. The dataset was introduced in an ACL 2021 paper and is intended for non-commercial research use.

Use Cases

• Train abstractive summarization models on the structured facet summaries to generate multi-perspective document condensations.
• Analyze the relationship between document sections and specific summary facets to understand information extraction patterns.
• Develop evaluation metrics for faceted summarization by comparing model outputs against the provided structured summaries.

Strengths

• Over 60,000 summary entries provide a substantial corpus for model training and evaluation.
• Data is structured for the specific research task of faceted summarization, as detailed in the ACL 2021 paper.
• The dataset is tagged for the summarization task and text modality, clarifying its intended application.

Limitations

• The dataset is restricted to non-commercial research use only, limiting its applicability for commercial projects.
• Access requires manual approval via email, which may delay or prevent immediate use.
• Specific details on document sources, column structure, and file formats are not provided in the input.

Provenance

Source

Emerald publications, as indicated by the description snippet 'Over 60k Emerald…'.

Collection Method

null

Time Range

null

Freshness

null

Geography

null