name: qa-reading
description: "Guides question-answering and reading-comprehension dataset
discovery in CLUEDatasetSearch, including DuReader, SQuAD, NewsQA, CMRC, CAIL,
cMedQA, CoQA, and search-log QA resources."
disable-model-invocation: true
metadata:
disco-role: operating
license: NO_LICENSE
QA and Reading Dataset Route
Use this sub-skill when the user intent matches one of these signals:
- question answering, open-domain QA, answer selection
- reading comprehension, cloze, yes/no, multi-document or legal RC
- DuReader, SQuAD, CMRC, CAIL, NewsQA, WikiQA, MS MARCO
Read references/dataset-selection.md for QA/RC comparisons and references/troubleshooting.md for duplicate names and access issues.
Fast workflow
- Restate the user's requested task, language, domain, and output need.
- Search the bundled root index before recommending a row:
python ../../scripts/search_dataset_index.py --query dureader
python ../../scripts/search_dataset_index.py --category reading-comprehension --query legal
python ../../scripts/search_dataset_index.py --query SQuAD --limit 0
- Compare title, category, description, keywords, provider, license, paper, and note fields.
- Warn that CLUEDatasetSearch is a catalogue: no dataset files are bundled and external access must be verified.
- If the request belongs to another family, route to the sibling sub-skill named below instead of forcing a weak match.
Boundary routes
Output pattern
When answering, include:
- candidate dataset title and category;
- why it fits the task;
- language/domain/scale hints from the catalogue;
- provider and source URL when present in the bundled index;
- license/access caveat, especially if blank or competition-hosted;
- what the user must verify upstream before download, redistribution, or benchmark use.
Do not tell future agents to open or run files from the original repository checkout.