Data Dictionary

Relational Schema Specification

INCIDB ships as a 4-table normalized relational schema exported into pipe-delimited UTF-8 CSV (`|`) and high-performance Apache Parquet (`pyarrow`) archives.

1. brands.csv / brands.parquet

Contains canonical brand identities and identities across 5,994 global cosmetic lines.

Field Name	Data Type	Description	Sample Value
`brand_id`	`INTEGER`	Primary key unique identifier	`1`
`name`	`STRING`	Standardized commercial brand name	`Laneige`

2. ingredients.csv / ingredients.parquet

Contains 44,816 canonical chemical compounds normalized from public regulatory and open-data sources.

Field Name	Data Type	Description	Sample Value
`ingredient_id`	`INTEGER`	Primary key unique identifier	`1`
`inci_name`	`STRING`	International Nomenclature Cosmetic Ingredient name	`DIISOSTEARYL MALATE`
`cas_number`	`STRING`	Chemical Abstracts Service registry number	`50-81-7`
`common_name`	`STRING`	Plain English common compound name	`Diisostearyl Malate`
`primary_function`	`STRING`	Functional category (CosIng taxonomy)	`Skin-Conditioning Agent`
`comedogenic_rating`	`FLOAT`	Pore-clogging likelihood scale (`0.0` - `5.0`)	`0.0`
`is_common_allergen`	`INTEGER`	FDA MoCRA contact dermatitis allergen (`1`/`0`)	`0`
`is_fungal_acne_trigger`	`INTEGER`	Malassezia folliculitis feeding trigger (`1`/`0`)	`0`
`description`	`STRING`	Plain-language functional summary	`Standard cosmetic emollient...`
`fda_warning`	`STRING`	US FDA OTC warning monograph	`No warning required`

3. products.csv / products.parquet

Contains 19,645 commercial cosmetic formulations harvested from prestige retailers, K-Beauty standards, and clinical registries.

Field Name	Data Type	Description	Sample Value
`product_id`	`INTEGER`	Primary key unique identifier	`67455`
`brand_id`	`INTEGER`	Foreign key referencing `brands.brand_id`	`46067`
`barcode_ean`	`STRING`	Universal GTIN / EAN-13 barcode	`4006381333931`
`name`	`STRING`	Full commercial product title	`Good Genes Lactic Acid Treatment`
`category`	`STRING`	Primary skincare taxonomy category	`Skincare`
`retail_price_usd`	`FLOAT`	Retail price in USD	`85.0`
`raw_ingredient_text`	`STRING`	Full unbroken ingredient listing from package label	`Botanical Blend [Water/Eau/Aqua...`
`created_at`	`STRING`	Ingestion timestamp (ISO 8601)	`2026-07-02T14:01:48Z`

4. product_ingredients.csv / product_ingredients.parquet

Relational junction table containing 323,730 mappings preserving exact label order and concentration index.

Field Name	Data Type	Description	Sample Value
`product_id`	`INTEGER`	Foreign key referencing `products.product_id`	`67455`
`ingredient_id`	`INTEGER`	Foreign key referencing `ingredients.ingredient_id`	`12404`
`position_index`	`INTEGER`	Exact label position index (`1` = highest concentration)	`1`
`concentration_percentage`	`FLOAT`	Explicit clinical concentration percentage (when disclosed)	`10.0`