Helllo,
We have a Neo4J-db with about 30.000 nodes which is the basis of a complete web-app (including user-management, invoicing etc...).
Many of the nodes (around 20.000) contain (legal) text in html-format as a property of the respective nodes.
How can we build a knowledge graph of these nodes without interfering the existing graph?
Is it possible to kind of prefix the label of the nodes and relationships which will be created by the LLM, so that there is no interference with existing nodes? (e.g. (:User) is a normal node in the db whereas (:KG_User) is a node representing a User in the knowledge graph).
Or do you recommend to have the knowledge graph separated from the graph containing the text (e.g. in different db-instances)? And if they are separated in 2 different instances, how could one link from the "knowledge graph" to the "document graph"?
THX, JJJ