WeKnora is a framework designed for complex document understanding scenarios. It integrates multimodal pre-processing, semantic retrieval, and LLM-based reasoning, using RAG to strengthen generation and supporting PDF, Word and image formats.
Key features
- Multiformat document parsing and knowledge extraction
- Built-in knowledge graph construction and visualization
- Flexible retriever strategies (BM25, dense, graph-based RAG)
- Web UI and standard RESTful APIs for fast integration