Discovery of global knowledge in a database for cooperative answering

This paper proposes an approach to discover global knowledge from a large number of data in a database. In this approach, linguistic labels are defined by fuzzy sets, and they are regarded as predefined background knowledge to discover global knowledge. Using the linguistic labels of the background knowledge, the given data are classified into several subclasses. Recursive classification of the process constructs a hierarchy of classes. However, in a database there are some exceptional data that are difficult to be classified into the classes. Therefore, we have introduced exceptional classes, and the exceptional data are grouped into them. Using this technique, we have constructed an intelligent data retrieval system that provides the user with cooperative answers. When too many data are retrieved, our system provides the user with a linguistic answer that summarizes the retrieved data. When no data is retrieved, our system provides the user with a linguistic summary of what kinds of data are stored in the database. >