An empirical study on Resource Description Framework reification for trustworthiness in knowledge graphs

Area of Science:

• Computer Science
• Data Science
• Information Science

Background:

• Knowledge graphs (KGs) represent web knowledge using Resource Description Framework (RDF) triples.
• Data quality in KGs is a concern due to potential errors, outdated, or conflicting information.
• Metadata, such as source, location, and time, is crucial for enhancing fact trustworthiness and contextualization.

Purpose of the Study:

• To provide an overview of prominent RDF reification approaches for metadata representation in KGs.
• To analyze the representation and extent of provenance and context data in Wikidata and YAGO4.
• To investigate how these popular KGs handle metadata for fact trustworthiness.

Main Methods:

• Overview of prominent RDF reification techniques.
• Analysis of metadata representation in Wikidata using qualifiers.
• Analysis of metadata collection in YAGO4, leveraging Wikidata qualifiers.

Main Results:

• Wikidata utilizes qualifiers to embed metadata into facts.
• YAGO4 aggregates metadata primarily from Wikidata qualifiers.
• Both Wikidata and YAGO4 allow fetching facts without reification, supporting applications not requiring metadata.

Conclusions:

• RDF reification, while common, increases data size and doesn't always ensure metadata annotation.
• Wikidata and YAGO4 offer distinct approaches to metadata integration, with YAGO4 building upon Wikidata's structure.
• This research is the first to comparatively investigate the metadata handling methods and coverage in Wikidata and YAGO4.