Babel Gardens API

LIVELLO 2 service for multilingual semantics: embeddings, sentiment/emotion signals, profiles, and semantic enrichment.

Location

• Service: services/api_babel_gardens/
• Pure core (LIVELLO 1): vitruvyan_core/core/cognitive/babel_gardens/

Base URL / Port

Default port is 8009 (env: PORT).

Endpoints (implemented)

Health / metrics

• GET /
• GET /health
• GET /metrics

Embeddings

Defined in services/api_babel_gardens/api/routes_embeddings.py:

• POST /v1/embeddings/create — embedding for a single text
• POST /v1/embeddings/multilingual — embedding with language-aware path
• POST /v1/embeddings/batch — batch embeddings
• POST /v1/embeddings/similarity — cosine similarity between two texts

Sentiment

Defined in services/api_babel_gardens/api/routes_sentiment.py:

• POST /v1/sentiment/analyze
• POST /v1/sentiment/batch

Emotion

Defined in services/api_babel_gardens/api/routes_emotion.py:

• POST /v1/emotion/detect

Profiles + cognitive bridge

Defined in services/api_babel_gardens/api/routes_admin.py:

• POST /v1/profiles/create
• POST /v1/profiles/adapt
• POST /v1/profiles/recommend
• POST /v1/events/publish
• POST /v1/routing/intelligent

Scope boundary (important)

• Babel Gardens is a semantic-linguistic service, not the graph router.
• It provides language/emotion/sentiment/embedding signals to callers.
• Execution routing is decided by LangGraph orchestration (intent_detection_node.py + route_node.py), not by Babel Gardens.
• Legacy api_semantic is removed; do not model this service as “api_semantic replacement for routing”.

Request/response schemas

Pydantic models live in services/api_babel_gardens/schemas/api_models.py.

Most important request types:

• EmbeddingRequest (text, language, model_type, use_cache)
• BatchEmbeddingRequest (texts[], language, model_type, use_cache)
• SentimentRequest (text, language, fusion_mode, use_cache)
• EmotionRequest (text, language, fusion_mode, use_cache)

Service config env vars

Loaded in services/api_babel_gardens/config.py:

• HOST, PORT, DEBUG, LOG_LEVEL
• EMBEDDING_SERVICE_URL, EMBEDDING_ENDPOINT, EMBEDDING_TIMEOUT, EMBEDDING_RETRIES
• SENTIMENT_SERVICE_URL, SENTIMENT_ENDPOINT, SENTIMENT_TIMEOUT
• QDRANT_HOST, QDRANT_PORT, QDRANT_COLLECTION_SEMANTIC, QDRANT_COLLECTION_SENTIMENT
• POSTGRES_HOST, POSTGRES_PORT, POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD
• REDIS_HOST, REDIS_PORT, REDIS_DB