Redis Query Engine Demo

Two Complementary Search Approaches: Traditional + Vector Demonstrate Redis's versatility as both a traditional and vector DB.

Two Demo Flows

Flow 1: Traditional Search Demo Keyword syntax

• Setup: Movies + RediSearch indexes
• Capabilities: Field matching, ranges, sorting
• Example: @genre:{Action} @rating:[8 +inf]
• Value: Redis as a powerful search database

Flow 2: Vector Search Demo

Embedding-based semantic search - Natural language with HNSW vectors

• Setup: Movies + embeddings + vector index
• Capabilities: Semantic + Hybrid search
• Examples:
  • Semantic: space adventure with aliens
  • Hybrid: superhero movie | genre:Action year>2010
• Value: Redis with vector semantic search!

Code Structure

• src/core/ - Redis config, embeddings, index schemas
• src/data/ - Data loading and index creation
• src/search/ - Traditional, semantic, and vector search
• src/utils/ - As the name implies

Quick Start

1. Setup

git clone <repo>
cd redis-movie-search
pip3 install -r requirements.txt

# Configure Redis connection
cp .env.example .env
# Edit .env with your Redis credentials

2. Seed Data + Setup

python3 run.py setup

3. Run the Demo

# Flow 1: Database syntax
python3 run.py search-basic

# Flow 2: Natural language  
python3 run.py search-advanced

Examples

Keyword Search (Flow 1)

Traditional keyword syntax with precise field matching:

@genre:{Action} @rating:[8 +inf]
@title:star wars
@release_year:[2010 2020] SORTBY rating DESC
@genre:{Comedy} @rating:[7 10] LIMIT 0 5

Hybrid Search (Flow 2)

Embedding-based semantic search with natural language queries:

Semantic (vector similarity):

space adventure with aliens
romantic comedy in Paris
movies about time travel
psychological thriller

Hybrid (vector + traditional filters):

superhero movie | genre:Action year>2010
comedy | rating>7.5 year>2015
space adventure | year<2000

Requirements

• Python 3.8+
• Redis Stack (RediSearch + Vector support)
• 1GB RAM for embeddings