The Design And Implementation Of An RDF Dataset Summarization Tool

Recently, the semantic web has been more and more popular due to its advantage on data querying, data fusion and reasoning. With the push of Linking Open Data project, a large number of RDF datasets have been published. Although those RDF datasets can provide a lot of information to people, it’s also a hard task to understand and use those datasets because of the large amounts and scale. "Summarization" presents the key content of a complex thing by using a simple and condensed version. With the help of RDF dataset summarization, those RDF datasets can be much easier to understand and use.However, there’s a lack of RDF dataset summarization tools.Therefore, the paper proposes a framework about how to summary an RDF dataset. The framework defines "the summarization generation of a RDF dataset" as a problem of "content representing, sorting and selecting".The framework summaries an RDF dataset mainly based on the salience of both schema level and instance level. The framework also considers the coverage to improve the result summarization. Based on the framework, the paper also proposes an RDF dataset summarization tool "SaIBDS(Schema and Instance Based Dataset Summarization)". Not just providing a fixed summarization result, SaIBDS provides more choices to users. Users can decide the ratio between schema and instance, the scale of the summary and the algorithms of salience computing. It’s more flexible and user-friendly. Because users can explore the summary from different facets, they can get a better understanding of the RDF dataset.The paper evaluates the summarization system by an experiment which collected and analyzed users’feedbacks and the summarization method by an experiment which compared the summarization result with a Golden Standard answer.The results of the experiments indicate that SaIBDS system can provide a relatively good RDF dataset summarization which presents the key content of the RDF dataset.