Improve vegan/plant-based product taxonomy and reduce categorization bias

[stuckvgn]

Problem

The current taxonomy system treats vegan and plant-based products as secondary classifications rather than first-class categories. This creates several issues for the growing plant-based consumer base:

1. Vegan/plant-based as labels, not categories: In taxonomies/labels.txt, there is no dedicated vegan or plant-based label entry — only a 100% vegetable claim exists. Meanwhile, taxonomies/ingredients_analysis.txt defines vegan/vegetarian status outputs (Vegan, Non-vegan, Maybe vegan, Vegan status unknown), but these function as computed attributes rather than browsable product categories.

2. Category hierarchy bias: Plant-based products are typically nested under their animal-product counterparts (e.g., Plant-based milks as a subcategory of dairy-adjacent categories). Consumers who default to plant-based encounter the animal-product framing first, with plant-based as an alternative.

3. Ingredient analysis defaults: When "milk" appears in a product like chocolate, the system may default toward assuming dairy unless explicitly overridden — even as oat milk chocolate, coconut milk chocolate, etc. become increasingly common.

Proposed solution

1. Elevate vegan/plant-based to first-class categories

Add top-level category entries in taxonomies/food/categories.txt:

• en:Plant-based foods (top-level, not nested under animal-product categories)
• en:Plant-based milks, en:Plant-based cheeses, en:Plant-based meats as peers to (not children of) their animal-product counterparts

2. Expand taxonomies/labels.txt with proper vegan/plant-based labels

Add structured label entries for:

• en:Vegan (with multilingual translations, wikidata links, proper hierarchy)
• en:Plant-based / en:100% plant-based
• en:Vegan Society certified, en:V-Label Vegan, etc. as children

3. Improve ingredient analysis for ambiguous ingredients

• When a product is explicitly labeled/tagged as vegan, do not flag plant-based versions of traditionally animal ingredients as ambiguous
• Add awareness that "milk" can refer to plant-based milks — especially when the product name or other ingredients confirm this
• Reduce false "Maybe vegan" / "Vegan status unknown" classifications

Additional context

At Open Paws (501(c)(3) nonprofit building AI infrastructure for animal advocacy), we work with 100+ advocacy organizations globally and would be happy to contribute taxonomy improvements including multilingual support and testing.

Code pointers

• taxonomies/food/categories.txt — main product category hierarchy
• taxonomies/labels.txt — product label/certification taxonomy
• taxonomies/ingredients_analysis.txt — vegan/vegetarian status classification
• taxonomies/food/ingredients.txt — ingredient definitions