Moderately than storing knowledge in rows and columns for conventional searches, or as embeddings for vector search, a data graph represents knowledge factors as nodes and edges. A node might be a definite truth or attribute, and edges will join all of the nodes which have related relationships to that truth. Within the instance of a product catalog, the nodes stands out as the particular person merchandise whereas the sides might be related traits that every of these merchandise possess, like dimension or shade.
Sending a question to a data graph includes on the lookout for all of the related entities to that search, after which making a data sub-graph that brings all these entities collectively. This retrieves the related data for the question, which might then be returned again to the LLM and used to construct the response. This implies which you could cope with the issue of getting a number of related knowledge sources. Moderately than treating every of those sources as distinct and retrieving the identical knowledge a number of instances, the info might be retrieved as soon as.
Utilizing a data graph with RAG
To make use of a data graph along with your RAG software, you may both use an present data graph with knowledge that’s examined and recognized to be right prematurely, or create your individual. When you find yourself utilizing your individual knowledge—reminiscent of your product catalog—it would be best to curate the info and examine that it’s correct.
