Product descriptions are embedded for similarity search and filtered by category and publication status.
Keep vector dimensions and metadata types consistent.
Which design is most appropriate?
Define a knn_vector mapping with a fixed dimension and separate structured fields for attribute filters.
Detailed explanation
A path string is not a structured search mapping and is difficult to update.
A path string is not a structured search mapping and is difficult to update.
The mapping defines vector dimensions, similarity behavior, and typed filters for combined retrieval.
The mapping defines vector dimensions, similarity behavior, and typed filters for combined retrieval.
TTL expires items and does not define index types or dimensions.
TTL expires items and does not define index types or dimensions.
A text field cannot perform numeric nearest-neighbor distance calculations safely.
A text field cannot perform numeric nearest-neighbor distance calculations safely.
Check official documentation
Compare the following points with the official documentation.
Amazon OpenSearch Service vector search、knn_vector、index mappingの公式説明を確認する。
What to confirm
ベクトル型・次元数・属性フィールドをmappingで管理する理由を説明できる。
Key points
- knn_vector
- Vector dimension
- Index mapping
- Attribute 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
Model compatibility
Changing an embedding model may change dimensions or distance behavior and require a new mapping.
Hybrid retrieval
Vector similarity can be combined with text and attribute constraints and then evaluated.