BASF-AI/ChemNQRetrieval

---
dataset_info:
- config_name: corpus
features:
- name: _id
dtype: string
- name: title
dtype: string
- name: text
dtype: string
splits:
- name: corpus
num_bytes: 11814444.868325485
num_examples: 22933
download_size: 6046630
dataset_size: 11814444.868325485
- config_name: default
features:
- name: query-id
dtype: string
- name: corpus-id
dtype: string
- name: score
dtype: float64
splits:
- name: test
num_bytes: 1110.7605332063795
num_examples: 35
download_size: 2196
dataset_size: 1110.7605332063795
- config_name: queries
features:
- name: _id
dtype: string
- name: text
dtype: string
splits:
- name: queries
num_bytes: 1724.433371958285
num_examples: 27
download_size: 2922
dataset_size: 1724.433371958285
configs:
- config_name: corpus
data_files:
- split: corpus
path: corpus/corpus-*
- config_name: default
data_files:
- split: test
path: data/test-*
- config_name: queries
data_files:
- split: queries
path: queries/queries-*
task_categories:
- question-answering
language:
- en
tags:
- chemistry
- wikipedia
- nq
- natural questions
- chemteb
pretty_name: Chemical Natural Questions
size_categories:
- 10K<n<100K
license: cc-by-nc-sa-4.0
---
# Chemical Natural Questions

This dataset is created from the [mteb/nq](https://huggingface.co/datasets/mteb/nq) dataset on Hugging Face, which is part of the [Natural Questions](https://ai.google.com/research/NaturalQuestions/) dataset containing real user questions issued to Google search, with answers sourced from Wikipedia. In this chemistry-specific subset, we filtered queries related to chemistry by starting from the chemistry category in Wikipedia and traversing up to three levels deep in linked articles. This approach allowed us to focus on chemistry-related queries, providing a targeted subset of the original dataset for domain-specific retrieval tasks.