Online recommendation in non-stationary environments based on knowledge graph enhancement and time-varying reward mechanism | Synapse