Skip to Content (Press Enter)
Community › Forge › Assets list › RAG Knowledge Base
Login to follow
RAG Knowledge Base

RAG Knowledge Base (ODC)

Stable version 0.1.14 (Compatible with ODC)
Uploaded on 21 Aug (12 days ago) by DB Results Labs
RAG Knowledge Base

RAG Knowledge Base (ODC)

Details
Detailed Description

The RAG Knowledge Base is a template app for OutSystems Developer Cloud. It gives you a working starting point for building document ingestion, semantic search, and hybrid retrieval pipelines natively in ODC, without an external vector database or search service.


The app includes pre-built screens, pre-configured entities, and a hybrid search pipeline that combines semantic vector search with BM25 lexical scoring, fused using Reciprocal Rank Fusion. Wire up your embedding API credentials, run the BM25 index timer, and start from a working foundation rather than from scratch.


Key Screens

File Upload: upload a PDF, chunk the text, generate embeddings, and build the BM25 index in one workflow.

Semantic Search: query in natural language, toggle between semantic and hybrid search, retrieve ranked chunks with page references.


Documents: inspect ingested documents and processing status.


Vectors: inspect stored chunks, vector entries, and BM25 token detail per chunk.

ODC Asset screenshot_0ODC Asset screenshot_1ODC Asset screenshot_2ODC Asset screenshot_3ODC Asset screenshot_4
Limitations
  • Dependencies required. Semantic Engine V2 Library, SemanticEngineV2 External Logic, and BM25Engine External Logic must be installed for ingestion and hybrid search to work.
  • Large PDFs may timeout. Ingestion runs synchronously on screen. Very large files can hit ODC's execution time limit. Run ingestion in a background Timer or ODC Workflow for large documents.
  • Linear scan does not scale indefinitely. Both vector search and BM25 scoring scan all stored chunks and posting rows at query time. This works well for small to medium knowledge bases. For large document collections, query time will degrade and an external vector database is the right tool.
  • BM25Stats must be initialised before hybrid search. RecomputeBM25Stats must run after the first index. If BM25Stats has no row, HybridSearchWithBM25 returns an error.
Release notes 

Static Entity "RetrievalMode" is made public

License

https://opensource.org/licenses/BSD-3-Clause