Workshop Proceedings TEXT2KG is the third Workshop on Knowledge Graph Generation From Text co-located with ESWC-2024. The primary aim of this workshop is to bring together researchers from multiple focus areas such as Natural Language Processing (NLP), Entity Linking (EL), Relation Extraction (RE), Knowledge Representation and Reasoning (KRR), Deep Learning (DL), Knowledge Base Construction (KBC), Semantic Web, Linked Data, and other related fields to foster a discussion and enhance the state-of-the-art in knowledge graph generation from the text. The workshop The organizing team is thankful to everyone involved in making the TEXT2KG workshop 2024 a success. First, our thanks go to all the organizers of the main events and Program Committee members for ensuring a rigorous review process that led to an excellent scientific program and an average of three reviews per article. TEXT2KG team is also thankful to keynote speaker Publicity Chair (Joey Yip and Ronak Panchal) for their valuable contributions.
CEUR
ceur-ws.org CEUR Workshop Proceedings • Knowledge Graphs for Digital Transformation Monitoring in Social Medi • Open Knowledge Base Canonicalization: Techniques and Challenges • Towards Dataset for Extracting Relations in the Climate-Change Domain • Leveraging Language Models for Generating Ontologies of Research Topics • On Constructing Biomedical Text-to-Graph Systems with Large Language Models • Towards LLM-driven Natural Language Generation based on SPARQL Queries and RDF
Knowledge Graphs • Incorporating Type Information Into Zero-Shot Relation Extraction • Towards Harnessing Large Language Models as Autonomous Agents for Semantic Triple
Extraction from Unstructured Text • Battalogy: Empowering Battery Data Management through Ontology-driven Knowledge
Graph
Keynote Paul Groth Talk Title ”To Graph or not to Graph: Knowledge Graph architectures and LLMs”
The increasing capabilities of Large Language Models (LLMs) allows us to rethink how we architect systems for and abased on knowledge graphs. LLMs can be used as encoders for unstructured information such and images and text, which allows to take advantage of the attributes of entities in a Knowledge Graph. Furthermore, LLMs provide increasingly robust information extractors allowing information to be extracted on-the-fly. Likewise, they can be used as flexible components of common data wrangling tasks such as entity resolution. Finally, LLMs contain knowledge in their parameters providing a new source of knowledge. Together these suggest the ability to create new architectures that can take advantage the of diferent characteristics of information sources.
• We have taken the opinion of all organizers and steering and advisory committee to decide the best paper award and it was finally awarded to: – Fine-Tuning vs. Prompting: Evaluating the Knowledge Graph Construction with LLMs. Hussam Ghanem and Christophe Cruz Organizing Committee
• Amit Sheth, University of South Carolina, USA • Alfio Gliozzo, IBM Research, USA • Sören Auer, Leibniz University of Hannover and TIB, Germany • Enrico Motta, The Open University, UK • Anna Fensel, Wageningen University & Research, The Netherlands, & University of
Innsbruck, Austria • Maria Esther Vidal, Leibniz University of Hannover and TIB, Germany • Edlira Vakaj, Birmigham City University, UK • Fernando Ortiz-Rodriguez, Universidad Autonoma de Tamaulipas, Mexico • Sven Groppe, University of Lübeck, Germany
• Hong Yung (Joey) Yip, AIISC, University of South Carolina, USA • Ronak Panchal, Cognizant, India
• Tommaso Soru, University of Leipzig, Germany
Acknowledgements The editors would like to wholeheartedly thank supporting keynote, advisory team, authors, the program committee and other organizers for their constant support to make this event successful.