Category: Ingestion & Data Management
Priority: Medium
Phase: Phase 2 (Ingestion + CLI)
Problem
Chunking strategy is unspecified per content type — fixed-size chunking would destroy slide structure and over-chunking papers loses argumentative context.
Solution
Define a chunking strategy per content type in rag_core (structure-aware for slides, hierarchical for papers), record as an ADR, and store the strategy as chunk metadata.
Category: Ingestion & Data Management
Priority: Medium
Phase: Phase 2 (Ingestion + CLI)
Problem
Chunking strategy is unspecified per content type — fixed-size chunking would destroy slide structure and over-chunking papers loses argumentative context.
Solution
Define a chunking strategy per content type in
rag_core(structure-aware for slides, hierarchical for papers), record as an ADR, and store the strategy as chunk metadata.