varunshenoy/GraphGPT
GraphGPT
GraphGPT es un generador de grafos de conocimiento basado en LLM que extrae entidades y relaciones de texto no estructurado para crear grafos de conocimiento visuales. Funciona como una interfaz de grafo de lenguaje natural y un pipeline de extracción de datos no estructurados, transformando texto sin procesar en triples estructurados para mapear redes de información complejas.
El sistema permite el mapeo dinámico de conocimiento al permitir a los usuarios construir y actualizar visualizaciones de red mediante consultas conversacionales e instrucciones basadas en texto. Esto permite convertir datos no estructurados en grafos visuales para identificar patrones y conexiones entre entidades.
La herramienta cubre la extracción de entidades de información y grafos de conocimiento, proporcionando un pipeline para convertir lenguaje natural en representaciones estructuradas de información conectada.
Features
- Knowledge Graph Extraction - Automatically identifies entities and relationships from text to build structured visual knowledge representations.
- Data Extraction Pipelines - Provides a pipeline for transforming raw natural language into structured triples for information mapping.
- Entity Extraction Pipelines - Uses LLMs to automatically parse raw text into structured nodes and edges for graph construction.
- Dynamic - Enables the dynamic expansion and updating of visual information networks through conversational queries.
- Natural Language Query Interfaces - Provides a natural language interface for building and updating network visualizations.
- Text-to-Graph Transformers - Transforms unstructured natural language into structured property graphs for visualization.
- Information Extraction - Extracts key entities and their relations from unstructured text for systematic analysis.
- State Continuity Injections - Implements iterative context injection to ensure continuity when incrementally building knowledge graphs.
- Node Graph Serializers - Serializes complete node graphs including positions and connections into JSON for visualization.
- Graph Property Modifiers - Allows modification of graph vertices and edges through natural language instructions.
- Graph-Based Visualizations - Converts plain text into visual graphs to make complex connections and patterns easier to understand.
- Graph Schema Extractions - Constrains extracted data to specific node and edge definitions based on a graph-specific data model.
- Incremental State Accumulation - Maintains a persistent knowledge graph state that is updated as user queries provide new information.
- General Utilities - Tool for converting natural language into structured knowledge graphs.
