|
|
--- |
|
|
pretty_name: TREC Question Classification |
|
|
task_categories: |
|
|
- text-classification |
|
|
language: |
|
|
- en |
|
|
configs: |
|
|
- config_name: default |
|
|
data_files: |
|
|
train: data/train.parquet |
|
|
validation: data/validation.parquet |
|
|
test: data/test.parquet |
|
|
--- |
|
|
|
|
|
# TREC |
|
|
|
|
|
A classic benchmark dataset for question classification with both coarse and fine-grained labels. |
|
|
|
|
|
- **Size:** small, clean, ready to use |
|
|
- **Source:** [original release](https://cogcomp.seas.upenn.edu/Data/QA/QC/) |
|
|
- **Format:** stored in Parquet |
|
|
- **Compatibility:** 🧩 works with `datasets >= 4.0` (script loaders deprecated) |
|
|
|
|
|
## Reference |
|
|
|
|
|
Li, X., & Roth, D. (2002). |
|
|
*Learning Question Classifiers.* |
|
|
[ACL Anthology](https://aclanthology.org/C02-1150/) |
|
|
|
