Vector Databases : A Practical Introduction.

Format:

Book

Author/Creator:

Borwankar, Nitin.

Language:

English

Subjects (All):

Databases.

Vector processing (Computer science).

Natural language generation (Computer science).

Generative artificial intelligence.

Physical Description:

1 online resource (293 p.)

Place of Publication:

Sebastopol : O'Reilly Media, Incorporated, 2026.

Summary:

The AI revolution is here, and at its core lies a game-changing technology that most developers haven't fully explored: vector databases.From powering semantic search to enabling large language models (LLMs) and generative AI, vector databases are reshaping how we build applications with unstructured data like text, images, and audio.

Contents:

Cover

Copyright

Table of Contents

Preface

What's in This Book

Who This Book Is For

How to Use This Book

Software, Environment, and Resource Requirements

Conventions Used in This Book

Using Code Examples

O'Reilly Online Learning

How to Contact Us

Acknowledgments

Chapter 1. Introduction to Vector Databases

Why Do You Need Vector Databases?

A New Data Type: Vector

Similarity Search

What's Different About the Vector Type?

Where Do You Use Vector Databases?

SQL Versus Vector Databases

The Foundation of Business Math: Accounting Arithmetic

Vector Representation in a Relational Database Management System

The Need for Vector-Specific Capabilities

NoSQL Versus Vector Databases

NoSQL Databases and Vector Storage

Limitations of Vector Extensions in NoSQL Databases

When to Choose NoSQL with Vector Extensions

Hybrid Approaches: Combining Structured and Vector Data

The Need for Both Vector Data and Metadata

Limitations of Pure Vector Storage

Hybrid Database Architecture

Example of a Hybrid Query

Benefits of the Hybrid Approach

Conclusion

Chapter 2. Embeddings

Understanding Vector Embeddings: Why We Need Them

Word2Vec: The Breakthrough That Changed Everything

Doc2Vec: From Words to Documents

From Embeddings to Modern Language Models: The Transformer Connection

Encoder-Only Transformers (BERT and Its Variants)

Decoder-Only Transformers (GPT Family)

Encoder-Decoder Transformers (T5, BART)

Embedding Models: The Specialized Vector Generators

Distinction from Traditional Models

Role in Modern LLM Applications

Practical Applications and Use Cases

Simple RAG Pipeline

The sentence-transformers Library: The Swiss Army Knife of Text Embeddings

Best Practices for Using SentenceTransformers: A Detailed Guide

The Embedding Layer: The Gateway to Zero-Shot Learning

Anatomy of Transformer Embeddings

Connection to Zero-Shot Learning

Key Characteristics That Enable Zero-Shot Learning

Limitations and Considerations

Latest Developments and Trends

Vector Arithmetic with Word2Vec: A Hands-On Guide

Step 1: Setup and Installation

Step 2: Load Pretrained Word2Vec Model

Step 3: Implement Vector Arithmetic Functions

Step 4: Classic King-Queen Analogy

Step 5: More Interesting Analogies

Step 6: Interactive Exploration Tool

Final Words on Vector Arithmetic

Chapter 3. Similarity Search with FAISS

Foundations

Vector Representations

Distance Metrics

Selection Heuristics

FAISS Indexes

Flat Indexes (Brute Force)

IVF-Based Indexes

LSH-Based Indexes

HNSW-Based Indexes

Other Specialized Indexes

Composite and Transformative Indexes

Choosing the Right Index

Quantization

SQ

PQ

The ANN Problem

The Problem

Avoid Computational Cost

Notes:

Description based upon print version of record.

Key ANN Techniques in FAISS

OCLC-licensed vendor bibliographic record.

ISBN:

1-0981-7758-4

OCLC:

1584459392