A method and apparatus for of high-speed and memory efficient rule matching, the rule matching being performed on an m-dimensional universe with each dimension bound by a given range of coordinate values, and a set of rules that apply to an undetermined number of coordinates in that universe. More specifically, a high-speed computer based packet classification system, uses an innovative set intersection memory configuration to provide efficient matching of packets flowing through a network system to a specific process flow based on a packet tuple. The system also provides classification of packets as they flow through a network system. More particularly, this system correlates these flowing packets with previously received packets, along with identifying the packets so that they are handled efficiently. The ability to correlate packets to their corresponding process flows permits the implementation of service aware networks (SAN) that are capable of handling network situations at the application level.