Evaluacion de la Calidad Linked Open Data basada en un Modelo de Lógica Difusa

Abstract

Linked Open Data has been one of the most widely used online data publishing methods in recent years. This growth means that the quality of this data is required for the benefit of consumers and people who wish to use this data. There are approaches based on classical mathematical models, however, most of these results are too linear; that is, they use conventional evaluators to define both quality aspects and results. In response, a new approach based on fuzzy logic is constructed as an application, which aims to complement and compare traditional models without the need to restrict the quality aspects with which it can be measured. As a methodology, it is done by obtaining data from each dataset through the SPARQL Endpoints provided by high category datasets, classifying them within accessibility and trust dimensions, represented in 4 values: response time, scalability, trustworthiness and timeliness. This analysis is done internally for the values within the accessibility dimension, and externally for the values within the confidence dimension. In this way, it is possible to know or determine a better general quality approximation of the Linked Open Data according to a large number of quality evaluation variables, or even parameterize its own aspects in the model as a complement to the already established models, through the concept of fuzzy logic.