Open-Assistant
Open-Assistant copied to clipboard
Add the CosIng dataset (cosmetic ingredients)
CosIng is an open EU dataset of cosmetic ingredients.
I think it's what websites like CosDNA or INCIdecoder are using with a few tweaks for better explainability. Link to dataset (CSV available): https://data.europa.eu/data/datasets/cosmetic-ingredient-database-ingredients-and-fragrance-inventory?locale=en
It roughly follows this structure:
INCI name |
Chem/IUPAC Name / Description |
Restriction |
Function |
|---|---|---|---|
| DIPTEROCARPUS INTRICATUS EXTRACT | (Dipterocarpus Intricatus Extract is the extract of the whole plant Dipterocarpus intricatus, Dipterocarpaceae. | ANTIOXIDANT, HUMECTANT, SKIN CONDITIONING, SKIN CONDITIONING - EMOLLIENT | |
| 1,2-BUTANEDIOL | HUMECTANT, SKIN CONDITIONING, SOLVENT, VISCOSITY CONTROLLING |
The dataset could be converted to QA / instructional using this format:
Q: What is the <INCI name>?
A: Here is the short description or chemical name of this ingredient: <Chem/IUPAC Name / Description>. It's used as: <Function>
And for rows with empty descriptions:
Q: What is the <INCI name>?
A: I couldn't find a description for this ingredient. However, it's used as: <Function>
What do you think?