Building Knowledge Graph from Relational Database
Bilal Ben Mahria, Ilham Chaker, Azeddine Zahi · 2023
Semantic-based data integration has grown to be one of the largest data challenges in the last 20 years, especially when the data is released in a variety of forms and uses a variety of schemas (e.g., HTML, CSV, JSON, spreadsheet, SQL data). As a result, using knowledge graph could efficiently help to address the problems related to data integration. The knowledge graph consists of a number of interconnected descriptions of entities (real-world objects as documents or abstract concepts such as a Person that is a being that has attributes like morality and consciousness, etc.) where these descriptions have formal semantics that makes it possible for both humans and machines to analyze them effectively and unambiguously. Ontologies– which can be thought of as the knowledge graph&s;s schema – are used to create a formal meaning for these entities. The ontologies work as a formal agreement that guarantees a common understanding of the data and its meaning between the creators of the knowledge graph and its consumers. The construction of ontology is an activity that is linked strongly with engineering. Therefore, to develop an ontology, there are two main approaches: (i) manual development (called also from scratch) or (ii) ontology learning development. Indeed, it is very hard and expensive to develop an ontology manually because the requirements call to combine the skills of domain experts’ knowledge and ontology engineers’ expertise. In this context, the term “ontology learning” has come into use, which describes a strategy for automatically or semi-automatically learning ontological knowledge from a structured, unstructured, or semi-structured source of data. In this chapter, we aim to provide a detailed review of state-of-the-art relational database to ontologies approaches by categorizing and comparing the methods. The methods involved in each approach will be discussed in terms of their capabilities, advantages, and drawbacks.