What Makes a Good Ontology? A Case-Study in Fine-Grained Knowledge Reuse

摘要

Understanding which ontology characteristics can predict a "good" quality ontology, is a core and ongoing task in the Semantic Web. In this paper, we provide our findings on which structural ontology characteristics are usually observed in high-quality ontologies. We obtain these findings through a task-based evaluation, where the task is the assessment of the correctness of semantic relations. This task is of increasing importance for a set of novel Semantic Web tools, which perform fine-grained knowledge reuse (i.e., they reuse only appropriate parts of a given ontology instead of the entire ontology). We conclude that, while structural ontology characteristics do not provide statistically significant information to ensure that an ontology is reliable ("good"), in general, richly populated ontologies, with higher depth and breadth variance are more likely to provide reliable semantic content.