Today, we are announcing the support of the open-source GraphRAG Toolkit, a new capability that enhances Generative AI applications by providing more comprehensive, relevant and explainable responses using RAG techniques combined with graph data. The toolkit provides an open-source framework for automating the construction of a graph from unstructured data, and composing question-answering strategies that query this graph when answering user questions.
Previously, customers faced challenges in conducting exhaustive, multi-step searches across disparate content. By identifying key entities across documents, GraphRAG delivers insights that leverage relationships within the data, enabling improved responses to end users. For example, financial analysts can ask a financial analysis chatbot for the sales forecast of a manufacturing company. Developers building Generative AI applications can enable GraphRAG via this new open-source Python toolkit by specifying their data sources and choosing Amazon Neptune Database or Neptune Analytics as their graph store and Amazon OpenSearch serverless as the vector store. This will automatically generate and store vector embeddings in the selected vector store, along with a graph representation of entities and their relationships in the selected graph store.
The GraphRAG Toolkit is an open source project. Its code base is open for inspection, modification, and extension, and is therefore highly adaptable for specific or niche requirements. With its initial release, the toolkit provides graph store implementations for both Neptune Analytics and Neptune Database, and vector store implementations for Neptune Analytics and OpenSearch Serverless, and it uses FMs hosted in Amazon Bedrock. To learn more, visit the User Guide
.