Everyday situations require us to reason. For solving complex problems in our real world, it is important not only to get explicit information from the outside world, but also to seek and identify appropriate information by selecting it from a huge amount of knowledge stored in memory. Therefore, reasoning is fundamental to human intelligence. The most important process of reasoning is to select optimal knowledge which is essential to interpretation of current information, and to ignore inappropriate knowledge which is irrelevant. In Relevance Theory, it is claimed that an optimal relevance gives the most appropriate interpretation by means of deductive inference. This paper provides a way of computational formalization of Relevance Theory and proposes a system of an interpretation of some aspects of Japanese conditionals.
展开▼