Building and managing fuzzy ontologies with heterogeneous linguistic information

Fuzzy ontologies allow the modeling of real world environments using fuzzy sets mathematical environment and linguistic modeling. Therefore, fuzzy ontologies become really useful when the information that is worked with is imprecise. This happens a lot in real world environments because humans are more used to think using imprecise nature words instead of numbers. Furthermore, there is a high amount of concepts that, because of their own nature, cannot be measured numerically. Moreover, due to the fact that linguistic information is extracted from different sources and is represented using different linguistic term sets, to deal with it can be problematic. In this paper, three different novel approaches that can help us to build and manage fuzzy ontologies using heterogeneous linguistic information are proposed. Advantages and drawbacks of all of the new proposed approaches are exposed. Thanks to the use of multi-granular fuzzy linguistic methods, information can be expressed using different linguistic term sets. Multi-granular fuzzy linguistic methods can also allow users to choose the linguistic term sets that they prefer to formulate their queries. In such a way, user-computer communication is improved since users feel more comfortable when using the system. Ontologies have become an important tool in the domain mod-eling field. Thanks to them, it is possible to carry out real world representations, stablish axioms and obtain conclusions of them [1,2]. Ontologies have been wide used in several fields. In biomedi-cine field [3–5], ontologies have been employed, for example, to build knowledge databases about genes and proteins characteristics that help researchers to classify and understand how the human body works. In semantic web field [6–8], ontologies have been used to classify concepts that can be referred through the web. This way, searches are improved and give better results to the users because a concept, instead of a word that can have different meanings, is used. In the artificial intelligence field [9–11], ontologies can also be applied to create knowledge databases to be used in systems that employ the provided information to carry out different tasks. However, classical Crisp Ontologies have one important drawback , that is, their element descriptions can only be expressed using crisp membership values. Consequently, each described element has a set of fulfilled characteristics and another one with characteristics that do not describe the element. That is, membership value of each element to each concept is represented by the values {0,1} where 0 means that the …