News Generating Based on Interval Type-2 Linguistic Summaries of Databases

Generating the so-called linguistic summaries of databases, exemplified by About half of records have very high values of attribute A, in sense of Yager [17] [19] with further improvements [2] [5] is described. The original authors’ contribution is the use of Interval Type-2 Fuzzy Sets [8] [9], instead of ordinary fuzzy sets, to model imprecise but understandable natural language propositions like high cost, small amount. It is necessary to obtain the knowledge representation which is not arbitrary, but build on opinions given by different experts and/or collected from different sources. The described methods provide the foundations for automated generating of textual news to be published in e-press, WWW, or in RSS channels. A sample implementation and its results are presented.