You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Neo4j added a new type of index in version 5.11 : Vector index
It comes with a set of commands to query it.
But it is still in beta mode, so the set of commands are subject to change, so I suggest we hold off a bit before adding support for this new index type ; but happy to debate.
The text was updated successfully, but these errors were encountered:
So, the _install_node / _install_relationship are the ones where the magic happens (if a node has an index property, it will create an index.
A few tricks here though :
We must pass an extra property to mention the type of index we want to create (so we create a vector and not the default range index)
We must check that the Neo4j version is right to create vector indexes. There are utility methods and examples for this, like : if self.version_is_higher_than("5.7")
An underlying question is that we should provide a way to make neomodel query using vector indexes, since it's a specific syntax in Cypher (something like CALL db.index.vector.queryNodes) and returns a list of elements with a similarity score. So that would require a separate method I guess ?
Neo4j added a new type of index in version 5.11 : Vector index
It comes with a set of commands to query it.
But it is still in beta mode, so the set of commands are subject to change, so I suggest we hold off a bit before adding support for this new index type ; but happy to debate.
The text was updated successfully, but these errors were encountered: