Search Databases¶

Flexible GraphRAG supports three full-text search options for the hybrid search pipeline.

BM25 (Built-in)¶

Local file-based BM25 full-text search with TF-IDF ranking. No external server required.

• Dashboard: None (file-based)
• Best for: Local development, simple deployments

SEARCH_DB=bm25
SEARCH_DB_CONFIG={"persist_dir": "./bm25_index"}

Elasticsearch¶

Enterprise search engine with advanced analyzers, faceted search, and real-time analytics.

• Dashboard: Kibana at http://localhost:5601
• Docker: Uncomment includes/elasticsearch-dev.yaml in docker-compose.yaml

SEARCH_DB=elasticsearch
SEARCH_DB_CONFIG={"hosts": ["http://localhost:9200"], "index_name": "hybrid_search"}

OpenSearch¶

AWS-led open-source fork with native hybrid scoring (vector + BM25) and k-NN algorithms.

• Dashboard: OpenSearch Dashboards at http://localhost:5601
• Docker: Uncomment includes/opensearch.yaml in docker-compose.yaml

SEARCH_DB=opensearch
SEARCH_DB_CONFIG={"hosts": ["http://localhost:9201"], "index_name": "hybrid_search"}

python scripts/create_opensearch_pipeline.py
# or: scripts/setup-opensearch-pipeline.sh  (Linux/macOS)
# or: scripts/setup-opensearch-pipeline.bat  (Windows)

Disabling Full-Text Search¶

To use vector search only (no full-text):

SEARCH_DB=none

Full Configuration Reference¶

See Database Configuration for the full search database configuration reference with all options.