Automated Class Correction and Enrichment in the Semantic Web

Abstract

The Semantic Web is an effort to interchange unstructured data over the Web into a structured format that is processable not only by human beings but also computers. The key backbones of Semantic Web are ontologies and annotations that provide semantics for data. Ontologies are usually created before actual data is populated. Subsequently, they can be incomplete and they often do not provide all aspects that are required for specific domains of knowledge. Additionally, Semantic Web-based ontologies usually suffer from a considerable amount of faulty facts which are known as Semantic Web data quality issues. Due to the complexity of relationships, Semantic Web data quality issues are continuously growing. This paper follows two main objectives. Firstly, it concentrates on a specific Semantic Web data quality issue that indicates incorrect assignment between instances and classes in the ontology. Secondly, the paper shows how to discover new classes which are not denied in the ontology and how to place them in the hierarchical structure of the ontology. To make ends meet, an entropy-based approach called ACE (Automated Class Corrector and Enricher) is proposed that not only evaluates the correctness and incorrectness of relationships between instances and classes but also generates new classes to enrich ontologies. The contributions of ACE have been also explained throughout the paper. Initial experiments conducted on a Semantic Web dataset demonstrate the effectiveness of the ACE.

Keywords: Semantic Web data quality issues, Ontology, Incorrect class assignment, Class enrichment, Ontology enrichment, Information theory