RDF graph pair profile dataset for the data linking community

Area of Science:

• Computer Science
• Data Science
• Semantic Web Technologies

Background:

• The increasing volume of RDF datasets necessitates effective methods for linking similar entities across diverse sources.
• Data linking is crucial for integrating and enhancing information within the Web of Data.

Purpose of the Study:

• To introduce the "RDF graph pair profiles dataset" to support the data linking community.
• To facilitate the development, selection, and validation of data linking tools through comprehensive dataset profiles.
• To provide a valuable resource for machine learning applications in data linking.

Main Methods:

• The dataset comprises profiles for 30 RDF graph pairs, categorized by ontology matching (OM), instance matching (IM), or both (OM + IM).
• Each profile includes statistical measures, qualitative and quantitative information, and descriptive models generated via automated tools.
• The methodology for profile generation is briefly outlined, and both original RDF graphs and their profiles are provided.

Main Results:

• The dataset offers detailed profiles that characterize RDF graphs, aiding in the understanding of their properties.
• Profiles serve as input parameters for machine learning models, enhancing their applicability in data linking tasks.
• The dataset is publicly available, promoting its use by researchers and practitioners.

Conclusions:

• The "RDF graph pair profiles dataset" is a comprehensive resource that will significantly facilitate data linking efforts.
• Its availability supports the integration and enhancement of RDF data on the Semantic Web.
• This work contributes to advancing the capabilities of data linking tools and methodologies.