BASF-AI/PubChemSMILESCanonDescBM

---
dataset_info:
features:
- name: description
dtype: string
- name: canonical_smiles
dtype: string
splits:
- name: test
num_bytes: 3000530
num_examples: 15457
download_size: 1353859
dataset_size: 3000530
configs:
- config_name: default
data_files:
- split: test
path: data/test-*
license: cc-by-nc-sa-4.0
language:
- en
tags:
- chemistry
- pubchem
- SMILES
- chemteb
pretty_name: PubChem Canonical SMILES and Descriptions Bitext Mining
size_categories:
- 10K<n<100K
---
# PubChem Canonical SMILES and Descriptions Bitext Mining
This dataset contains two separate lists: one of canonical SMILES strings and the other of corresponding entity descriptions, both sourced from PubChem ([ChEBI](https://www.ebi.ac.uk/chebi/) source). The task is to identify matching pairs between the SMILES strings and the descriptions, where each SMILES string from the first list should be aligned with its corresponding description from the second list. The dataset is intended for bitext mining tasks, where the goal is to correctly retrieve the matching description for each chemical entity represented by its SMILES string, providing a valuable resource for chemical entity alignment and retrieval tasks.