The latest release of neo4j/neo4j-graphrag-python is 1.22.0. This repository is the official library for building graph retrieval-augmented generation (GraphRAG) applications with Neo4j and Python. However, the collected information contains no changes for 1.22.0. Only the release number is confirmed, and there is no way to know what this version changed. The source of the information is the neo4j/neo4j-graphrag-python repository on GitHub.
What this package does is provide a path to build a knowledge graph and retrieve from it. It covers document-based graph construction together with the combination of vector search and graph search: feeding text and PDF files into Pipeline or SimpleKGPipeline builds a knowledge graph. Vector index creation and similarity search through VectorRetriever, as well as GraphRAG queries, are handled inside the same library. The repository has 1316 GitHub stars, the primary language is Python, and the creation date is February 27, 2024.
The license, however, is listed as NOASSERTION, so no license name identified by GitHub is given. From this information alone it is not possible to judge which usage permissions apply. No performance figures or benchmarks are presented either, so in the 10-point evaluation only the momentum item remains at 1.0 while the novelty, technology, and quality items are empty. Since neither the changelog nor the license terms are confirmed, it is premature to use 1.22.0 as the basis for an adoption decision.