This section is about how memories are structured as data — schemas, entities, facts, provenance, and graphs — so recall stays precise as knowledge grows and changes.
What you will learn
- How to model memories as collections, objects, and properties
- When to use structured facts versus free-text narratives
- How knowledge graphs and temporal facts complement vector search
- Versioning, merging, confidence, and contradiction at the representation layer
Chapters in this section
There are 16 chapters in this part. Open any chapter to read it on its own, or work through them in order.
- 1 Collections, Objects, and Properties: Structuring Memory as Data
- 2 Schema Design for Memory Systems
- 3 Structured Memories vs Free-Text Memories
- 4 Entities and Relationships in Agent Memory
- 5 Knowledge Graphs as a Complement to Vector Search
- 6 Temporal Knowledge Graphs: Modeling Facts That Change
- 7 Fact Extraction: Turning Conversation Into Structured Memory
- 8 Atomic Facts: Why Granularity Matters
- 9 Deduplication at the Representation Level
- 10 Property-Based Metadata for Memory Filtering
- 11 Modeling Confidence and Provenance in Stored Memories
- 12 Linking Memories to Source Data
- 13 Versioning Memories Over Time
- 14 Object Identity and Memory Merging
- 15 Cross-Referencing Related Memories
- 16 Representing Uncertainty and Contradiction in Memory