Internal documents will support question answering with stored embeddings.
Provide semantic retrieval and source references, and re-ingest documents when they change.
Which design is most appropriate?
Use Bedrock Knowledge Bases with a supported vector store and design chunking, metadata filters, and synchronization.
Detailed explanation
Lifecycle rules manage retention and transitions, not embeddings or retrieval.
Lifecycle rules manage retention and transitions, not embeddings or retrieval.
The flow can manage chunking, embeddings, vector search, and references to source documents.
The flow can manage chunking, embeddings, vector search, and references to source documents.
Audit events alone do not provide the requested document knowledge source.
Audit events alone do not provide the requested document knowledge source.
A full scan is not semantic nearest-neighbor search and grows in latency and cost.
A full scan is not semantic nearest-neighbor search and grows in latency and cost.
Check official documentation
Compare the following points with the official documentation.
Amazon Bedrock Knowledge Bases、vector store、embedding、metadata filteringの公式説明を確認する。
What to confirm
ベクトルインデックスと全文Scan・S3 Lifecycle・監査ログの役割を区別できる。
Key points
- Embedding
- Vector index
- Knowledge Bases
- Metadata filter
Notes
- Environment: AWS公式DEA-C01試験ガイドと各サービスの公式ドキュメントの確認
- Confirm current AWS service behavior in the official DEA-C01 documentation.
Foundation review
Chunking
Chunk boundaries and overlap affect retrieval quality and the context supplied to a model.
Synchronization
Updates and deletions in source documents require controlled re-ingestion and stale-vector cleanup.