In an inference machine having a knowledge set portion which stores therein a set of knowledge items such as rules, and an inference portion which interprets the rules for execution; priority level information items which set precedence degrees among the rules within the rule set are stored in a plurality of series with respect to standpoints of users, one series which meets the standpoint of a particular user is selected from among the plurality of stored series of priority level information items, and the rule of the highest precedence degree is selected on the basis of the selected series of priority level information items, with the result that the rule conforming to the standpoint of the particular user is selected and executed. On the other hand, a plurality of optimum solutions for the respective standpoints can also be found in such a way that the different series of priority level information items are successively selected, whereupon the inference portion finds the solutions for the respective selected series of priority level information items.