Incremental Knowledge Graphs

With knowledge graphs (KGs) at the center of numerous applications such as recommender systems and question answering, there is an increasing need for generalized pipelines that can not only construct but also continuously update these KGs.
While the individual steps necessary to create KGs from unstructured (e.g., text) and structured data sources (e.g., databases) are well-researched for one-time execution, their adoption for incremental KG updates and the systematic interplay of these steps have not been thoroughly investigated.
Addressing these gaps requires developing methodologies to efficiently handle updates within KG pipelines, creating frameworks for automatic pipeline configuration, and establishing benchmarking approaches to evaluate and compare different KG pipelines.
By focusing on these incremental aspects, we aim to ensure the adaptability, accuracy, and scalability of KG systems.

Aims

The project aims to develop a methodology for incorporating and handling updates efficiently within a Knowledge Graph (KG) pipeline, ensuring the system remains adaptive and up-to-date with minimal manual intervention.
Additionally, it seeks to create an Automatic Pipeline Configuration Framework that streamlines the setup and management of KG pipelines, enhancing ease of use and reducing configuration errors.
Furthermore, the project will establish a comprehensive Knowledge Graph Pipeline Benchmarking Approach to systematically evaluate and compare the performance, accuracy, and scalability of different KG pipelines.

Team

• Marvin Hofer
• Prof. Dr. Erhard Rahm

Publications

• Marvin Hofer et al. The new dbpedia release cycle: Increasing agility and
  efficiency in knowledge extraction workflows. In 16th International Confer-
  ence on Semantic Systems, SEMANTiCS, pages 1–18. Springer, 2020.
• Marvin Hofer, Daniel Obraczka, Alieh Saeedi, Hanna Köpcke, and Erhard
  Rahm. Construction of Knowledge Graphs: State and Challenges. CoRR,
  abs/2302.11509, 2023.
• Marvin Hofer, Johannes Frey, and Erhard Rahm. Towards self-configuring
  Knowledge Graph Construction Pipelines using LLMs – A Case Study with
  RML. In Fifth International Workshop on KGC @ESWC, 2024.