OpenSPG
OpenSPG is an open-source, semantic-enhanced programmable knowledge-graph engine developed by Ant Group in collaboration with the OpenKG community. It is intended to connect heterogeneous enterprise data, represent domain semantics and rules, and bidirectionally enhance large-language-model applications with controllable knowledge-graph construction, retrieval and reasoning.
At a Glance
- Enterprise data and AI teams
- Financial services and risk-management organizations
- Government and e-government service providers
- Healthcare and medical-information applications
- +2 more
AI Tools by OpenSPG
(1)
KAG
Knowledge Graph LLM Q&A Framework
Discussions
No discussions yet
Be the first to start a discussion about OpenSPG
Latest News
OpenSPG 0.8 released; release highlights include configurable indexes, public/private knowledge bases, MCP and HTTP/front-end integration, and KAG-Thinker support.
OpenSPG/KAG 0.7.1 released with compilation, diagnostics, scheduling, stability, streaming-output and monitoring improvements.
KAG 0.7 released with hybrid static/iterative planning, extensible symbolic solvers, MCP integration, lightweight construction and open benchmarks.
OpenSPG 0.6 released with expanded domain-knowledge mounting and related engine updates.
Products & Services
Apache-2.0 open-source knowledge-graph engine based on the SPG framework. It provides schema semantic modeling, structured and unstructured knowledge construction, programmable logical-rule reasoning, pluggable engine/algorithm adaptation and cloud adaptation components.
A logical-form-guided reasoning and retrieval framework built on OpenSPG and LLMs for professional-domain knowledge bases, supporting knowledge/chunk mutual indexing, semantic knowledge alignment, schema-constrained construction, hybrid reasoning and multi-hop factual Q&A.
An interactive thinking/deep-reasoning model integrated into KAG 0.8, supporting broad decomposition, deep solving, knowledge-boundary determination and noise-resistant retrieval for complex problems.
A related OpenSPG project described by the OpenSPG GitHub organization as a large-model-based knowledge-extraction framework with generalized Chinese and English, multi-field and multi-task extraction capabilities.
Market Position
OpenSPG positions itself between conventional enterprise knowledge-graph engines and LLM-native retrieval systems: unlike a basic property-graph database it exposes domain semantics, programmable rules and construction/reasoning operators, while unlike vector RAG or GraphRAG it emphasizes schema-constrained modeling, semantic alignment, logical-form-guided retrieval and deterministic reasoning. The KAG work compares against RAG-family and graph-RAG approaches for multi-hop and professional-domain QA.
Founding Story
OpenSPG grew out of Ant Group's Semantic-enhanced Programmable Graph (SPG) framework, which accumulated experience from financial-domain knowledge-graph deployments. Ant Group and OpenKG then jointly released the framework to provide an extensible open engine for domain knowledge modeling, construction and reasoning, and to connect knowledge graphs with large language models.
Business Model
Revenue Model
Open-source software distributed under the Apache License 2.0; the public project provides downloadable code and documentation rather than a publicly documented paid subscription or API pricing model.
Target Markets
- Enterprise data and AI teams
- Financial services and risk-management organizations
- Government and e-government service providers
- Healthcare and medical-information applications
- Researchers and developers building knowledge-graph, RAG and LLM reasoning systems
- Universities, laboratories and open-source contributors
- Enterprise supply-chain knowledge graphs and risk mining
- Financial risk warning, anti-black-market analysis and institutional risk assessment
- Event/causal graphs and macroeconomic, industry and corporate-event analysis
- Medical knowledge graphs and health question answering
- E-government question answering
- Professional-domain logical reasoning, multi-hop factual Q&A and knowledge decision support
- Ant Group e-government Q&A
- Ant Group e-health Q&A