I'm looking into a persistant representation of a naive Bayesian classifier using a graph database. I have three basic object types: users, words and and topics. The relationships between these nodes would represent the strength of their connection -- a probability between zero and one.
To query the graph I would traverse relationships from user to topic, using the strength of connections to represent connectedness. Querying could potentially take a more neural net-like form. I'm still quite naive myself when it comes to graph databases, but a Bayesian classifier seems to be a good fit for a graph model like Neo4j. That said, in my background research I haven't seen a way to represent the strength of connections, just the binary relationship of whether two objects are connected or not. Can anyone comment on the feasibility of a Neo4j implementation of a Bayesian classifier? Are there ways I might be able to represent relationship strength using Neo4j primitives? -- View this message in context: http://neo4j-community-discussions.438527.n3.nabble.com/Representing-relationship-strength-tp3352296p3352296.html Sent from the Neo4j Community Discussions mailing list archive at Nabble.com. _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
