Help Needed: RAG for Sparse SQL Databases
Desperate-Vast-4899 · reddit · 2026-07-10
The author is building a RAG pipeline using Ollama's Qwen2.5, aiming to let the LLM read a risk register database via SQL and semantic search to answer questions about risks, incidents, and mitigation measures. They are encountering the same core issue as a previous post: the database is highly sparse with many empty tables and columns, leading to a lack of retrieval context and unstable model answers. So far, the author has only implemented row-level chunking and embedding, and has yet to introduce solutions like hybrid search or RRF.
Related event: Help Needed: Building RAG Pipelines on Sparse SQL Databases(2 posts)→
More from coding & agent
- Codex turns out 123 screensavers in one playful batch — intellectronica · 2026-07-21
- Grok Build adds `grok doctor`, resumable sessions and remote image paste — mark_k · 2026-07-21
- Autoresearch proposes packaging ML runs as studies with questions, analysis, and code diffs — morgymcg · 2026-07-21
- CHAP defines approvals, handoffs, and audit logs for human-agent workflows — DeliveryTechnical199 · 2026-07-21
- The author says Codex reached 20x and is now debugging spec decoding on a hybrid parallel setup — TheZachMueller · 2026-07-21
- Axcess adds an MCP connector for WCAG accessibility checks that scanners miss — modelcontextprotocol · 2026-07-21