paper-reader

Extract structured evidence from abstracts by identifying task, method, dataset, supervision, and limitation details to support literature screening workflows. Output a structured dataset containing section, excerpt, note, and confidence fields for each evidence entry.

When full text is unavailable, evidence extraction proceeds using only the abstract with clear notes indicating the limitation and conservative confidence scores. The output explicitly flags missing full text so researchers can gauge the reliability of the extracted evidence.

You can extract methods and datasets from local PDFs by providing the paper metadata and full text to the skill. It processes the available content to output structured evidence entries covering tasks, supervision, and limitations required for your literature review.

The best way to get limitation details from an ambiguous abstract is to supply the full text for structured evidence extraction. The skill analyzes the complete document to isolate method, dataset, supervision, and limitation specifics that the abstract alone does not clearly state.

Evidence extraction works with only abstract text by applying conservative confidence scores and notes indicating full text is unavailable. It parses the available abstract to capture available task, method, dataset, supervision, and limitation information for your screening process.

The structured evidence dataset output includes fields for section, excerpt, note, and confidence. These fields organize the extracted task, method, dataset, supervision, and limitation details to integrate directly into your literature screening workflow.