metadata
license: apache-2.0
language:
- en
license: apache-2.0
task_categories:
- text-classification
- token-classification
tags:
- cosmetics
- llm
- ontology
- classification
- multilabel-classification
- machine-learning
- nlp
- hierarchical-classification
- cosmetic-chemistry
size_categories:
- 1K<n<10K
CosFunc-HO: A Hierarchical Cosmetic Ingredient Function Dataset
Overview
CosFunc-HO is a cleaned and structured cosmetic ingredient function dataset developed from COSDNA data for research in Large Language Models (LLMs), ontology learning, multi-label classification, and cosmetic chemistry analysis.
The dataset contains more than 9,000 labeled cosmetic ingredient records and includes hierarchical functional annotations organized into multiple levels of abstraction.
This dataset was created as part of a research study on evaluating Large Language Models for cosmetic ingredient function prediction using hierarchical ontologies.
Dataset Contents
Main Files
cosfunc_final_dataset_v4_1.csvcosfunc_final_ontology_v4_1.csv
Features
The dataset includes:
- Ingredient names
- Parent functional labels
- High-level functional domains
- Ontology mappings
Applications
This dataset can be used for:
- Multi-label classification
- LLM evaluation
- Ontology learning
- Hierarchical classification
- Cosmetic ingredient analysis
- Domain-specific NLP research
Domains Included
- Conditioning Agents
- Surfactants and Cleansing Agents
- Protective and Preservation Agents
- Formulation and Texture Modifiers
- Sensory and Aesthetic Agents
- Solvents and Carriers
- Mechanical and Physical Agents
- Hair Treatment and Styling
Research Context
This dataset supports research on:
- Large Language Model evaluation
- Hallucination analysis
- Hierarchical ontology design
- Structured prediction tasks
- Semantic label alignment
License
Apache License 2.0