FalkorDB reposted this
New DAGWorks Inc. post on Hamilton Burr FalkorDB for #GraphRAG! Graph DBs have always been really exciting technology -- does anyone else remember the release of Meta's graph-search? That said, they've had some adoption challenges. Generalizing (obviously), they (1) often require data to be extremely well structured for ingestion (relying on relationships that might not be known in advance or be subjective), and (2) often need a specific query language that was close-to-but-not-quite SQL. With the prevalence of #LLMs, however, I think that's all about to change. AI can help turn unstructured data into more structured data (deriving relationships), and turn natural language queries into graph queries. Generally I'm skeptical (bordering on cynical) of the GenAI craze, but this is something I think can really help move things forward. Stoked that Stefan Krawczyk had the chance to work with Falkor to show a very powerful example of how to do this, using both Hamilton *and* Burr. This post shows a full example of a highly parallelized ingestion ETL, a conversational interface, and use of the cypher query language (generated by an LLM) to build a powerful conversational interface on graph data. https://lnkd.in/g-WcJ5ku
