A Type-2 Fuzzy Approach to Linguistic Summarization of Data

This paper introduces an application of type-2 fuzzy sets in data linguistic summarization. The original approach by Yager (1982) based on representing natural language statements via type-1, i.e., the Zadeh fuzzy sets, is generalized with type-2 fuzzy sets applied as models of linguistically expressed quantities and/or properties of objects. Type-2 sets extend the known summarization procedures by handling fuzzy values stored in databases, and allow to represent a linguistic term via a few different membership functions (e.g., provided by different experts), which makes the method more general and human-consistent. Furthermore, quality measures for type-2 summaries are discussed in order to evaluate the informativeness of the messages generated. Finally, two prototype applications are presented and the success of the new method is discussed.