Researchers address a weakness in graph-based retrieval-augmented generation systems, where automatically constructed knowledge graphs often suffer from thematic irrelevance, logical inconsistencies, and structural fragmentation. MemGraphRAG uses a collaborative group of AI agents supported by shared memory to maintain a global perspective during graph construction, through unified schema filtering, conflict detection, and memory-guided bridging. The system achieved 59.25% average accuracy across multiple benchmarks, surpassing prior state-of-the-art baselines while retrieving results in an average of 0.061 seconds per query.
