As RFID technology has been widely adopted, it has been acknowledged that user-specific data carrying RFID tags can be conveniently used. Since the typical user-specific data is sensor data, some existing studies have proposed new RFID architectures for sensor data handling. However, there are industrial demands to include non-sensor data, such as sales records, repair history, etc., as well as sensor data in the purview of our study. To realize handling of such general user-specific data, we need to satisfy requirements such as the minimum set of data semantics required to access the user-specific data, flexible user-specific data memory schema determination and partial collection of user-specific data. For this purpose, we designed the role of "session manager", which communicates with associate applications and notifies an interrogator what to do upon retrieval of a unique identification number. The session manager with the structured user-specific data can provide flexible collection and processing of user-specific data in the networked RFID. Since the fundamental drawback of the session manager might be the lookup overhead, we examined the overhead with a simulation based on experimental data. It was revealed that the overhead is equivalent to and in some cases even better than that of the existing method, which relies on structural tag memory.