Update Graphrag from Microsoft: Improve Search by Dynamic Choice Communities

Microsoft has recently announced an important updating of its Graphrag system, aimed at improving the operation of search engines based on artificial intelligence. This update greatly accelerates the processing of large language models (LLM) and increases their accuracy. The updated version of Graphrag does not have a specific version number, but it is so significant that it is useful to call it 2.0 to distinguish from the original graphrag.

“Baseline rag struggles with queries that require aggregation of informing the ACross the Dataset to Compos and Answer. Queries Such as“ What Are Themes in the Data? ” Perform Terribly Because Baseline Rag Rags on A Vector Search of Semantically Similar Text Content for Within We Canver Such Questions, Because the Structure of the LLM-Generated Knowledge Graph Tells US About the Structure (and Thus themes) Into meningful semantic clusters that are pre-Summarized.

🚀Graphrag uses a two -step process to perform its work. In the first stage, it segments the search index for thematic communities that are formed around related topics. These communities are interconnected by means of entities (such as people, places or concepts) and relationships between them, forming a hierarchical knowledge. In the second stage, Graphrag uses the knowledge graph he created to give LLM context so that he can answer the question more accurately.

• 📌 Graphrag updates includes a "dynamic choice of communities" that evaluates the compliance of each community report. Inappropriate reports and their subordinates are removed, which improves efficiency and accuracy, focusing only on relevant information.