An adequate knowledge representation of time is a key ingredient in making intelligent computer programs more useful. However, such a representation should reflect how people usually view time. People tend to view time as having different topological properties under different circumstances. One such representation captures a cyclic view of time. I have designed a representation of cyclic time, which can be used to represent English-language utterances exhibiting such a conception of time in a knowledge base. This representation, a unique slot-and-filler-type frame structure, can capture the cyclic time knowledge elements needed to represent such utterances in a knowledge base. To this end I have extended Hart's (2000) overlay structure in several ways.; The development of this frame structure, which contains eight knowledge elements, is the result of the collection, analysis, and categorization of 185 natural language utterances exhibiting a cyclic notion of time. Examples come from many different sources such as newspapers, books, television, movies, verbal conversations, etc., to ensure that the sampling is representative of such a notion of time.; The taxonomy resulting from this analysis contains seven unique categories. Each category describes either some aspect of the frequency of the indicated cycle or a different characteristic of the indicated cycle itself—without regard to its frequency. This taxonomy is comprehensive because it can classify 175 out of 185 utterances.; To demonstrate the applicability of this cyclic time frame structure to areas of artificial intelligence, I have provided a detailed description of how it might be used in a security company's intelligent natural language query system. The description includes the major subsystems, program flow, and the hand-processing of several natural language queries.
展开▼