Class VectorDocument<T>

Namespace

AiDotNet.RetrievalAugmentedGeneration.Models

Assembly

AiDotNet.dll

Represents a document paired with its vector embedding for storage and retrieval.

public class VectorDocument<T>

Type Parameters

T

The numeric data type used for the vector embedding (typically float or double).

Inheritance

object

VectorDocument<T>

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

A VectorDocument combines a Document with its vector embedding, creating a complete unit ready for indexing in a vector store. The vector embedding captures the semantic meaning of the document's content in a numerical form suitable for similarity calculations.

For Beginners: A VectorDocument is like a book with its catalog card.

Think of it as two pieces working together:

• Document: The actual book (content, title, author, etc.)
• Embedding: The numerical "fingerprint" describing what the book is about

Why combine them? When you add documents to a search system, you need both:

• The vector (for finding similar documents through math)
• The document (for returning the actual content to users)

For example:

• Document: "Climate change affects global temperatures..."
• Embedding: [0.23, -0.45, 0.78, ..., 0.12] (768 numbers)

The system uses the numbers to search, then returns the text.

Constructors

VectorDocument()

Initializes a new instance of the VectorDocument class.

public VectorDocument()

VectorDocument(Document<T>, Vector<T>)

Initializes a new instance of the VectorDocument class with a document and embedding.

public VectorDocument(Document<T> document, Vector<T> embedding)

Parameters

document Document<T>

The document containing content and metadata.

embedding Vector<T>

The vector embedding of the document.

Properties

Document

Gets or sets the document containing the text content and metadata.

public Document<T> Document { get; set; }

Property Value

Document<T>

Remarks

For Beginners: This is the actual document with all its information. Contains the text, ID, metadata - everything except the vector.

Embedding

Gets or sets the vector embedding representing the document's semantic meaning.

public Vector<T> Embedding { get; set; }

Property Value

Vector<T>

Remarks

The embedding is a dense vector that encodes the document's content into a numerical representation. The vector dimension must match the embedding model's output dimension. Embeddings enable efficient similarity search through mathematical distance calculations.

For Beginners: This is the numerical "fingerprint" of the document.

Think of it like a GPS coordinate:

• Just as GPS uses (latitude, longitude) to represent a location
• An embedding uses hundreds of numbers to represent meaning
• Documents with similar meanings have similar numbers (close GPS coordinates)
• Different meanings have different numbers (far apart coordinates)

For example, embeddings for:

• "cat" and "kitten" would be close together (similar meaning)
• "cat" and "democracy" would be far apart (different meaning)