A non-transitory computer readable medium having a method represented by computer readable programming code encoded thereon, the method comprising the steps of:
- searching a case base for a user-validated knowledge statement having antecedent features that match each context feature of a user-specified context comprising more than one context features, the user-specified context received from a computer input device, wherein the case base is stored, along with a generalization/analog base, in at least one memory storage device in a computer, wherein the case base comprises user-validated knowledge statements and the generalization/analog base comprises computer-generated knowledge statements, each of the user-validated knowledge statements and the computer-generated knowledge statements comprising a consequent and an antecedent having more than one antecedent feature;
- causing a processor within the computer to display the user-validated knowledge statement consequent to the user via a display device if a user-validated knowledge statement having antecedent features that match the context features is found;
- and
- receiving, via the computer input device, user feedback regarding the displayed user-validated knowledge statement consequent, wherein if the user feedback indicates that the user-validated knowledge statement consequent is incorrect the method further comprises the step of creating, during runtime, a new rule by performing an analogical contextual transformation by
- selecting one context feature from the user-specified context,
- selecting, by chance in proportion to the size of the base, either the case base or the generalization/analog base,
- searching the selected base for a first knowledge statement having an antecedent feature that matches the context feature,
- searching the selected base for a second knowledge statement having a consequent that matches the consequent of the first knowledge statement,
- substituting an antecedent feature of the second knowledge statement with the selected context feature, and
- searching the selected base for a knowledge statement including antecedent features matching the other context features from the user-specified context and the selected context feature.