Combining Knowledge Graph and LLM to increase the factual accuracy of question-answering system
- kg_save: saves the pickle format knowledge graph object
- kg_visualization: saves HTML format knowledge graph visualization
- vdb / subgraph_vdb: vector_database for knowledge graph
- utils: All the helper functions
- gpt.py: wrapper of GPT
- kg_gen.py: All the functions for generating a knowledge graph
- similarity.py: cosine similarity function
- vdb.py: class and functions for managing a vector database
- KnowledgeGraph.py: the knowledge graph class
- main.py: entry point of constructing a knowledge graph
- qa.py: entry point of using knowledge graph to answer a question
- source: text being used to constructing a knowledge graph