Implement Retrieval module (Read Path) for KM-RAG Architecture #48

Closed
opened 2026-05-20 18:22:41 +00:00 by Antigravity · 0 comments
Collaborator

We need to implement the KM-RAG Retrieval Engine (Read Path) to allow users to ask complex questions, retrieve relevant semantic contexts from Qdrant, expand context via Neo4j conceptual relationships, and synthesize a grounded response with proper citations using Google Gemini (via Microsoft.Extensions.AI).

Scope of Work

  • Define query schema and DTOs (GroundedResponseDto, CitationDto).
  • Implement AskQuestionAsync method on IKnowledgeService/KnowledgeService using Qdrant, Neo4j, PostgreSQL, and Google Gemini.
  • Expose MediatR handlers and endpoints (/api/knowledge/ask, /api/knowledge/search).
  • Build a dedicated Blazor UI page (/intelligence) for testing and global Q&A with rich, premium aesthetics and micro-animations.
  • Write robust unit tests.
We need to implement the KM-RAG Retrieval Engine (Read Path) to allow users to ask complex questions, retrieve relevant semantic contexts from Qdrant, expand context via Neo4j conceptual relationships, and synthesize a grounded response with proper citations using Google Gemini (via Microsoft.Extensions.AI). ### Scope of Work - Define query schema and DTOs (GroundedResponseDto, CitationDto). - Implement AskQuestionAsync method on IKnowledgeService/KnowledgeService using Qdrant, Neo4j, PostgreSQL, and Google Gemini. - Expose MediatR handlers and endpoints (/api/knowledge/ask, /api/knowledge/search). - Build a dedicated Blazor UI page (/intelligence) for testing and global Q&A with rich, premium aesthetics and micro-animations. - Write robust unit tests.
Sign in to join this conversation.
No Label
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: mjasin/Nexus.Reader#48