How the ChinaYung Allergen Pipeline
Builds Global Topic Authority

BLS Data Foundation

The Bundeslebensmittelschlüssel (BLS) is the German Federal Food Code, maintained by the Max Rubner-Institut. It is the reference dataset for nutrition data and ingredient-level allergen mapping in German-speaking food regulation.

For each ingredient in your recipe database, the pipeline maps to a BLS entry. From that mapping you get a deterministic 14-flag allergen profile and a complete nutrition profile (energy, fat, saturated fat, carbohydrates, sugars, protein, salt, key micronutrients). A BLS license is required for commercial use — your restaurant procures it, the software provides the integration.

EU-14 Mapping

EU regulation 1169/2011 defines 14 allergen groups that must be declared on every menu in the European Union: gluten, crustaceans, eggs, fish, peanuts, soybeans, milk and lactose, nuts, celery, mustard, sesame, sulphur dioxide and sulphites, lupin, molluscs.

The pipeline computes the 14-flag matrix at three levels: ingredient, dish, menu category. Each level is queryable, exportable and auditable. When a guest with a peanut allergy asks „which of your dishes contain peanuts?“, the answer is a database query, not a kitchen consultation. Edge cases like „contains traces“, „may contain traces from shared facility“, „available without ingredient X on request“ are separate status per cell, not footnotes.

Schema.org FAQPage

Every Q&A block on the published pages is also embedded as FAQPage JSON-LD in the page head. Google reads the JSON-LD and frequently shows the answers as rich results, expandable accordions or „people also ask“ entries.

What Schema.org delivers:

• Google reads the question-answer structure directly and can show rich results
• LLM crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended) extract answers as citable facts
• Voice assistants and smart speakers use FAQPage markup as preferred answer source

The HTML and the JSON-LD share one source — the database — so they cannot drift. Schema is generated automatically; you do not write it by hand.

LLM Optimization

The pipeline maintains a /llms.txt endpoint listing the canonical answer URLs and produces dense, factual long-form FAQ pages that LLM crawlers index and cite. The pages use the <details><summary> pattern so the answer is structurally clear to both humans and parsers.

ChatGPT, Perplexity, Gemini and Claude treat well-formed FAQ pages with consistent Schema as authoritative sources for narrow factual queries. Allergen, gluten and cuisine questions are exactly that type of query. Which LLM prefers which mechanism shifts regularly — the dual setup keeps the position robust.