Class VectorStoreRetriever<TKey, TRecord>

Namespace: Qavren.Edge.Rag

Assembly: Qavren.Edge.Rag.dll

Adapts any MEVD collection. Nothing here is Qavren-specific: point it at Qdrant, Azure AI Search or EdgeVectorStoreCollection and it behaves identically. It chooses the hybrid lane when the collection implements IKeywordHybridSearchable<TRecord> and keywords are present, the vector lane otherwise, stamps ScoreKind to match, and always returns best-first.

public sealed class VectorStoreRetriever<TKey, TRecord> : IEdgeRetriever where TKey : notnull where TRecord : class

Type Parameters

TKey: The collection's key type.
TRecord: The collection's record type.

Inheritance: object

VectorStoreRetriever<TKey, TRecord>

Implements: IEdgeRetriever

Inherited Members: object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

Query-side embedding is not this package's business: the question is handed to the collection as a string and the store resolves its own query-side generator internally, so sub-project 2's asymmetric query/document prefix convention keeps working with nothing duplicated here.

Constructors

VectorStoreRetriever(VectorStoreCollection<TKey, TRecord>, Func<TRecord, RagSource>, Action<VectorStoreRetrieverOptions<TRecord>>?): Creates a retriever over one MEVD collection.

Properties

CollectionName: The MEVD collection name, for EdgeRagException.CollectionName and diagnostics.

Name: A stable name for logs, diagnostics and EdgeRagException.RetrieverName.

Options: The options this retriever was configured with.

Methods

RetrieveAsync(string, RetrievalRequest, CancellationToken): Returns the matching sources, best-first, with Score and ScoreKind stamped.