In this thesis, an efficient inference model for the domain independent expert system is presented. Conventional expert systems have used forward chaining control structure or backward one as their control structures. However forward chaining control systems generally have time-consuming firing strategy and cannot reason backward because of its restrictions of control structure. Backward chaining control systems cannot apply all the rules to the large set of the facts at once. The proposed model employs both two control structures and gives many facilities for the domain experts to save much time in problem solving. The proposed model, called PROBES(PROlog Based Expert System) is designed and implemented in PROLOG, a logic programming language.