Desenvolvimento de novas técnicas de extração de conjuntos de similaridade para relações não simétricas

Abstract

The main mathematical foundation of database management systems is the set theory. In a set, there are no pairs of equal elements. However, the exact comparison between complex data does not provide relevant information, and it is preferable to use comparisons by similarity. The concept of similarity sets was created to represent the idea of a set where there are no pairs of sufficiently similar elements. The theoretical basis of the similarity sets concept has been extended to address asymmetric similarity relations in this research. A new technique for the extraction of similarity sets, formally defined as algorithm Asymmetric Distinct, was developed and validated in two experiments. The technique was considered stable and scalable and allows for new research based on this study.