A More Specific Events Classification to Improve Crawling Techniques

Abstract

Nowadays the popularity of data quality is increasing notably in linked data. Linked data consuming applications need to be aware that changes in a dataset. Changes such as update, remove or creation links may occur for a time so is necessary to detect them to update local data dependencies where this annotation is made by detecting changes systems. Updated or removed links can be detected using a syntactic change similarity measure, and it can be done simply using the Levenshtein distance measure. However, a specific event subclassification of updated event and removed event, which iscreated by detecting changes systems developed, does not exist based on content analysis. A semantic signature and Maximum Similarity Measure (MaSiMe) combination approach is developed to create a more specific subclassification of the initial updated and removed event when its meaning has been changed. It is used to enrich the resources, annotating the new subclassification of the initial updated event and removed event, and will be annotated the author who created this annotation, adding provenance information. Annotations on the modification time are made in linked data resource, and making an average time study about when these specific events changes, could be improved the crawling techniques for a domain.