Knowledge graphs (KGs) are increasingly utilized in both private (e.g., Google) and public (e.g., DBpedia, Wikidata) domains, supporting various AI tasks like entity recognition and question answering. As Large Language Models (LLMs) gain prominence, KGs serve as reliable inputs for LLMs across user prompts, necessitating their construction and maintenance to ensure up-to-date knowledge for effective support.
This tutorial will impart attendees with standardized practices for declaratively constructing KGs from diverse input datasets. Using a real-world case study from the EU transport domain, participants will understand the importance of good practices in knowledge management. By the tutorial’s end, attendees will know how to establish sustainable workflows transforming data from any format (e.g., NLP-extracted entities in JSON) into well-formed RDF KGs. Key concepts to be learned by the attendees include the RDF Mapping Language (RML), virtual and materialized KGs, or advanced guidelines for complex and intense knowledge-driven environments.
