VSR - Verifiable Social Recommendation

VSR: Verifiable Social Recommendation

A decentralized recommendation system combining Zero-Knowledge Proofs and LLMs to enable privacy-preserving social recommendations.

Project description

VSR (Verifiable Social Recommendation) is an innovative system that combines Zero-Knowledge Proof (ZKP) technology with Large Language Models (LLMs) to create a decentralized recommendation platform. Unlike traditional centralized recommendation systems where companies control user behavioral data, VSR allows users to own their data while selectively sharing it for recommendations through a distributed approach.

The system addresses two key challenges:

1. Privacy & Trust: Uses ZKPs to prove data authenticity without revealing sensitive information
2. Cold Start Problem: Leverages LLM-powered AI agents to generate initial recommendations when user data is sparse

The architecture consists of two main components:

ZKP Core Module: A browser extension that extracts behavioral data from web services and creates verifiable proofs using Reclaim Protocol's @reclaimprotocol/zk-fetch. For this hackathon, we implemented Steam game purchase verification, allowing users to cryptographically prove they own specific games without revealing their credentials or complete purchase history.

Recommendation Application: A social platform where users can request game recommendations based on their verified Steam library. The system combines AI-generated suggestions from specialized agents with community-driven recommendations from verified game owners, creating a trustworthy and spam-resistant recommendation ecosystem.

Technologies used

• Frontend: React, Chrome Extension APIs
• Backend: Node.js, Express
• Zero-Knowledge Proofs: Reclaim Protocol (@reclaimprotocol/zk-fetch, @reclaimprotocol/js-sdk)
• AI/LLM: Large Language Models for recommendation generation
• APIs: Steam Web API for user data verification
• Verification: zkTLS for proving HTTP responses without revealing sensitive data

Basic architecture

Data Flow & System Components:

1. User Layer: Chrome Extension installed in user's browser extracts behavioral data (Steam cookies, user session) from logged-in Steam account without exposing credentials.
2. Verification Layer: Backend server receives user data and generates zero-knowledge proofs using Reclaim Protocol's zk-fetch. The system proves that specific conditions are met (e.g., user owns a particular game) without revealing the full purchase history or authentication details.
3. Proof Generation Process: The backend makes HTTP requests to Steam's user data API using provided session information, then creates cryptographic proofs that validate game ownership against specified conditions using regex pattern matching.
4. Recommendation Engine: Two-tier system combining AI agents specialized in different game genres with community-submitted recommendations from verified users. The LLM agents help solve cold-start problems by generating initial recommendations.
5. Social Platform: Web application where users can request recommendations, view AI and community suggestions, and submit their own recommendations. All user-submitted content includes verification badges showing proven game ownership.
6. Verification & Trust: Each recommendation displays whether the contributor actually owns the recommended game, creating a spam-resistant environment where only verified owners can make authoritative suggestions.

Source code

• https://github.com/TideQuest/vsr

Deployment

• Demo Platform: http://vsr-demo.tidequest.net/
• Chrome Extension: Available for manual installation (developer mode)

Setup Instructions:

1. Access the demo site at the provided URL
2. For ZKP verification, install the Chrome extension following the provided setup guide
3. Log into Steam in your browser before using the verification features

Key Features Available in Demo:

• Request game recommendations based on preferences
• View AI-generated and community recommendations
• See verified ownership badges on user recommendations
• Submit your own game recommendations with ownership verification
• Like and provide feedback on recommendations

Applications & Future Potential

The VSR framework extends beyond gaming to various domains:

• E-commerce: "Amazon purchase verification for exclusive access"
• Streaming: "Netflix viewing history for content-gated communities"
• Social Media: "Verified experience-based commenting systems"
• Education: "Course completion verification for advanced content access"
• Identity & Rewards: "Mizuhiki Verified SBT gating  submissions, trusted reviews, and exclusive campaign participation"

This creates new possibilities for trust-based digital experiences while preserving user privacy and data ownership.