Architecture for the Automatic Creation and Enrichment of Ontologies from Linked Data

Abstract

In smart environments (e.g.: smart cities) there are very diverse problems that must be addressed immediately; therefore, these environments must generate knowledge that allows them to respond to those particular needs, taking advantage of information from the context (e.g.: historical data, sensor measurement, problem description, among others). As a proposal for the timely generation of these knowledge bases, in this work an architecture is developed for allows to create and enrich emerging ontologies autonomously, using as input the linked data paradigm, which are data structures that are linked to each other. to serve both human users and other systems within the semantic web. This work specifies the services that offer the ability to generate emerging ontologies according to the problem that arises, seeking to learn the different axioms (concepts or properties) of the structures that the linked data sources used have. Then, these created ontologies are enriched/populated with the information that is extracted from the different linked data sources, to be returned to the requesting system automatically. For the development of this work, the MEDAWEDE methodology is used, which allows specifying this architecture, and conducting the construction of services to consume the sources described as linked data. Its six stages are used from this methodology: i) Specification: to analyze and select the ontological data sources to use. ii) Modeling: to implement the knowledge model of the study area of the ontologies to be generated. iii) Generation: to carry out the process of transformation, filtering, and integration of the data in the generated ontology. iv) Linking: to associate the data from the generated ontology with the knowledge from the linked data sets. v) Publication: to make the generated ontology available in the different standard formats. The exploitation stage is considered because it is dedicated to defining interfaces to access the ontology. The article also presents a case study focused on showing the process of generating an emerging ontology in the area of the Covid-19 pandemic, where the information available on the Internet is used through pre-existing ontologies and linked data sources.