LOD Link Discovery - Learning-based scalable link discovery for the data web

• Applicants: Professor Dr. Axel-Cyrille Ngonga Ngomo; Professor Dr.-Ing. Erhard Rahm
• Subject Area: Security and Dependability, Operating-, Communication- and Distributed Systems
• Term: from 2012 to 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 210434127

The cooperative project develops and evaluates advanced approaches for effective and scalable link discovery in the Data Web. We aim at an improved semantic linking of data sources as an essential prerequisite for data integration and other applications of Linked Open Data (LOD). In the proposed second phase of the project we want to investigate context-based and holistic link discovery approaches which utilize existing relationships and linkage results and which are not limited to linking only two data sources. To this end, we want to utilize and extend our previously developed web repository LinkLion providing many links between data sources. We also develop additional learning-based approaches for a simplified configuration of link strategies, in particular methods of statistical relational learning. Furthermore, we will comprehensively analyze the quality of computed links and improve or repair them by various methods. For high efficiency and scalability we develop approaches for parallel link discovery on graphics processors (GPUs) as well as on Hadoop clusters.

• DFG Programme: Research Grants