About WeKnora
WeKnora is an open-source LLM-powered framework developed by Tencent for deep document understanding, semantic retrieval, and context-aware question-answering. Built on the RAG (Retrieval-Augmented Generation) paradigm, WeKnora provides a comprehensive solution for transforming documents into intelligent, queryable knowledge bases.
Our Mission
WeKnora aims to democratize access to advanced document understanding and retrieval capabilities. Our mission is to provide organizations of all sizes with the tools they need to unlock the value hidden in their documents, enabling intelligent question-answering systems that understand context and provide accurate, relevant answers.
Project History
WeKnora was developed by Tencent as part of their commitment to open-source innovation. The framework emerged from the need to create a production-ready solution for document understanding and intelligent retrieval that could scale to enterprise needs while remaining accessible to developers and organizations of all sizes.
The project has grown to become a comprehensive framework with over 12,000 stars on GitHub, demonstrating its value to the developer community. WeKnora serves as the core technology framework for the WeChat Dialog Open Platform, powering intelligent Q&A services within the WeChat ecosystem.
Architecture Overview
WeKnora is built with a modern, scalable architecture that separates concerns and enables flexible deployment:
Core Components
- Document Reader: Advanced document parsing engine that supports multiple formats (PDF, Word, Markdown, etc.)
- Vector Database: High-performance vector storage and retrieval system for semantic search
- LLM Integration: Flexible integration with various LLM providers (OpenAI, Ollama, etc.)
- Reranking Engine: Advanced reranking algorithms to improve retrieval precision
- Multi-tenant System: Built-in support for multiple organizations and knowledge bases
- Web UI: Modern Vue.js-based interface for managing knowledge bases and conversations
- API Server: RESTful API built with Go for programmatic access
- MCP Server: Model Context Protocol server for extended tool integration
Technology Stack
WeKnora leverages modern technologies and best practices:
Backend
- Go (Golang) - High-performance backend services
- PostgreSQL - Relational database
- Vector Database - Semantic search storage
- Docker - Containerization
Frontend
- Vue.js - Modern reactive framework
- TypeScript - Type-safe development
- Modern CSS - Responsive design
AI & ML
- LLM Integration - Multiple provider support
- Embeddings - Vector representations
- Reranking - Advanced retrieval algorithms
- Agent Framework - ReACT pattern support
Infrastructure
- Docker Compose - Service orchestration
- Kubernetes (Helm) - Production deployment
- Jaeger - Distributed tracing
- MCP Protocol - Tool integration
Key Capabilities
Document Understanding
WeKnora can parse and understand documents in various formats, extracting structured information, identifying key concepts, and building semantic representations. The system automatically identifies document structures and extracts core knowledge to establish indexes.
Semantic Retrieval
Unlike traditional keyword-based search, WeKnora uses semantic retrieval to find relevant information based on meaning. This enables more accurate and contextually relevant search results, even when the exact keywords don't match.
Intelligent Q&A
WeKnora combines retrieval with generation to provide context-aware answers. The system retrieves relevant document sections and uses LLMs to generate comprehensive, accurate responses to user questions.
Knowledge Graph
Documents can be transformed into knowledge graphs, displaying relationships between different sections. This not only helps users understand document content but also provides structured support for indexing and retrieval.
Use Cases
WeKnora is being used in various scenarios:
- Enterprise Knowledge Management: Building intelligent knowledge bases for organizations
- Customer Support: Creating AI-powered support chatbots
- WeChat Integration: Powering Q&A services in the WeChat ecosystem
- Research & Documentation: Making research papers and technical docs searchable
- FAQ Systems: Building intelligent FAQ knowledge bases
Open Source Commitment
WeKnora is released under the MIT License, making it free to use, modify, and distribute. We believe in the power of open-source collaboration and welcome contributions from the community.
Our commitment to open source means:
- Full source code availability on GitHub
- Active maintenance and regular updates
- Community-driven development
- Transparent development process
- Free for commercial and non-commercial use
Project Statistics
WeKnora has gained significant traction in the open-source community:
12.2k+
GitHub Stars
1.4k+
Forks
26+
Contributors
624+
Commits
Get Started
Ready to start using WeKnora? Check out our getting started guide to set up your first knowledge base.