Modeling and linguistic knowledge extraction from systems using fuzzy relational models

Fuzzy relational models have been widely investigated and found to be an efficient tool for the identification of complex systems. However, little attention has been given to the linguistic interpretation of these models. The use of relational models is recommended since their development follows a natural sequence based on the original ideas about fuzzy sets and fuzzy logic, involving the estimation of the relations existing between linguistic terms which have previously been defined by the user. In the present paper the problem of extracting linguistic knowledge from systems by using relational models is addressed. A new algorithm for the identification of these models which can provide analytical or numerical solutions depending on user requirements is also proposed. Examples are presented showing that both quantitative and qualitative modeling can be effectively achieved by combining the proposed methodologies for identification and extraction of linguistic knowledge from systems.