This study proposes in the method for the prospective semantic inference in the web documents, catching the relation among RDF/RDFS resources, a core technique of the semantic web. For the semantic extraction, at first, we derive various kinds of facts caught intuitively from RDF/RDFS recommendation of W3C and stored them as facts in a knowledge base. Secondly, we produce rules from the various definitions of the recommendation and stored them as rules in a knowledge base, too. In this study. we choose the translation to the prolog, because the prolog is similar to the FOL representation. If there is a prolog compiler, we can easily confirm the inference process with the knowledge base. Rules and facts which represent the relation among the elements of the web documents will be helpful for semantic inference, provide the foundation available to agents.
