The engine that turns data into knowledge.
At Barliva's core is a knowledge graph, not a spreadsheet. The engine resolves messy, multi-source records into canonical entities and connects them so food can be queried, compared and reasoned about.
Canonical entities
Every raw record is resolved to a single source of truth: one product, one ingredient, one additive, one allergen, one brand — no matter how many ways they appear across sources or languages.
- Entity resolution & deduplication
- Confidence scoring with audit trails
- Conflict reconciliation across sources
- Stable identifiers you can build on
What the engine produces.
AI normalization
Cleans, parses and translates free-text fields into structured, queryable data.
Ingredient classification
Identifies ingredients, additives and allergens, mapping additives to E-numbers.
Multilingual resolution
Collapses language variants into one canonical entity — "tree nuts", "fruits à coque", "kuruyemişler".
Relations & graph queries
Connects entities so you can traverse products → ingredients → additives → risks.
Provenance
Every field traces back to its source for transparency and trust.
Continuous learning
New sources and corrections re-enrich entities over time.
Access the engine via API.
Bring canonical food intelligence into your own product.