Abordagem baseada na arquitetura DIKW para construção de grafos de conhecimento semântico que integram dados IoT com fontes de dados heterogêneas

Abstract

In recent years, Semantic Knowledge Graphs (SKGs) have gained a solid foothold in the market as a dynamic and scalable solution to meet the demands of large-scale data integration. In this context, an SKG functions as an information hub, encompassing data, metadata, and content, providing a unified, consistent, and unambiguous view of data from various sources and systems. However, building a semantic knowledge graph is a complex task that requires a meticulous approach to ensure its effectiveness, scalability, and ability to meet application objectives. Therefore, this work proposes an approach based on the DIKW architecture for the construction and specification of Semantic Knowledge Graphs that integrate IoT data with heterogeneous data sources. In this context, we investigate three fundamental research questions that result in three significant contributions: Question 1: How to build an IoT data SKG that is compatible with the different levels of the DIKW architecture? Question 2: How to build an SKG that integrates IoT data with heterogeneous data sources? Question 3: How to describe an IoT data knowledge graph, making it accessible and understandable to users? In response to the first research question, this work introduces the “DIKW4IoT” approach, which provides the necessary framework to represent IoT information at different granularities, allowing data to evolve within the DIKW hierarchy. This empowers them to transform into valuable insights and ultimately trigger informed actions and decisions. Regarding the second question, we propose “SIGIoT-S”, a conceptual framework that guides which data can be used in the integration process and identifies the necessary steps to achieve seamless integration. Finally, concerning the third question, we have developed the “MetaGraphIoT” ontology. This ontology aims to annotate metadata in an IoT data knowledge graph, making the knowledge graph more interpretable and usable for users. These contributions address the identified research questions and represent significant advancements in IoT data management, promoting better utilization and understanding of this data in various applications.