Page 125 - Intelligent Communication Systems
P. 125

104     INTELLIGENT COMMUNICATION SYSTEMS
        connects the call to Mr. Tanaka. With intelligent inquiry service, the system, when
        asked, say, "Please let me know of a cut-rate clothing shop in San Francisco," sup-
        plies the requested phone number.
            With  all  these  services,  the  system interprets  the inquiry from  a client and
        returns an answer. There  are a variety of requirements from clients,  expressed  in
        a variety of utterances. It is difficult  for a computer to understand every such utter-
        ance. Predicate  logic is one way to handle these problems,  and predicate  logic is
        appropriate not only for human beings but also for computers. The representations
        of knowledge, such as production rales and semantic networks, and the inference
        machine, also useful  for telecommunications, are described in this chapter, as are
        prepositional logic and predicate logic.




        10.1 APPLICATION OF PRODUCTION RULES
            TO TELECOMMUNICATIONS

        Production rules are used to describe the maintenance and operation of a communi-
        cation system. A production  rule (PR) is defined  as follows:

                          IF conditional clause, THEN action clause.

            An execution  algorithm  for a PR is shown in Figure  10.1. Here's a  strategy
        for  executing a production rule:

            (1)  A production rale that is matched to a variable in working memory
               (WM) is chosen.
            (2)  A production rale with the longest conditional  clause is chosen.
            (3)  All of the production rales that are matched to the variables in WM are
               chosen.
            (4)  All of the production rales that are matched to the variables in WM
               except the production rales that have been chosen  so far are selected.

           Example  1—Execution of Production Rules:
            (1)  In Figure  10.2,  "Pork" and "Cabbage" are in WM.
            (2)  Production rales PI  and P5, which are matched to "Pork" and
               "Cabbage," are chosen.
            (3)  P5 is selected  first  and executed. After execution,  "Vegetable"  is
               derived and stored in WM.
            (4)  P6, which is matched to "Vegetable," is chosen.
            (5)  P6 is executed. After  execution,  "Ted dislikes" is stored in WM. No
               production rale matched to "Ted dislikes" is found,  so execution is
               complete and the conclusion is that "Ted dislikes  cabbage."
            (6)  PI is executed and "Meat" is stored in WM.
   120   121   122   123   124   125   126   127   128   129   130