How sushant1209/hybrid-homomorphic-encryption-for-privacy-preserving-search works
Project type and purpose
• This is a command-line research/benchmark tool (not a web app) that generates synthetic encrypted datasets, runs two encrypted keyword-search algorithms (SCIE-Enc baseline vs. Ultra-Optimized Hybrid PHE+SwHE), measures performance, and produces comparison charts.
Main architectural components
a. Data Generation Module
• Script: generate_metadata.py
• Outputs: JSON files (metadata_profiles.json, metadata_scie.json, metadata_hybrid.json) in execution01/
b. SCIE-Enc Baseline Algorithm
• Built into ultra_optimized_hybrid.py under “SCIE-Enc” branch
• Reads metadata_scie.json, applies binary vector encryption & simple overlap matching
c. Ultra-Optimized Hybrid PHE+SwHE Algorithm
• Also in ultra_optimized_hybrid.py under “Hybrid” branch
• Sub-modules:
• PHE (Paillier addition)
• SwHE (optimized boolean AND/OR)
• Parallel processing engine (4 threads)
• Caching layer (lookup tables)
• Pre-computed encrypted constants
• Reads metadata_hybrid.json & metadata_profiles.json
• Produces ultra_optimized_results.csv
d. Data Storage / Results Folder
• Directory: execution01/
• Contains raw JSON inputs, CSV results, and generated graph PNGs
e. Visualization Module
• Script: create_graphs.py
• Reads ultra_optimized_results.csv
• Generates execution_time_comparison.png & match_count_comparison.png
Relationships and data flows
1. generate_metadata.py → execution01/*.json
2. ultra_optimized_hybrid.py reads JSONs → executes both algorithms in parallel → writes results CSV
3. create_graphs.py reads CSV → writes PNG charts
4. (CI) .github/workflows/run-project.yml automates full run
Architectural patterns & design principles
• Modular CLI steps: data-generation → computation → visualization
• Pipeline pattern: clear stepwise flow of data transformations
• Parallel batch processing for performance
• Separation of concerns: algorithms and visualization decoupled
• Caching & pre-computation for optimized hot-path operations
Technologies & dependencies
• Python 3.8+
• Homomorphic encryption library (e.g., Python-Paillier)
• numpy, multiprocessing, json, csv
• matplotlib or seaborn for plotting
Additional notes for clarity
• CI workflow as an orchestration layer that triggers all scripts in sequence
[Parallel Engine] → [PHE Module]
→ [SwHE Module]