A Pattern-based Annotation Transformation Schema for Knowledge Exchange

In teamwork of semantic annotation, the task is usually assigned to different annotators, and the process may last for a long period of time due to the complexity of its nature and coverage. Moreover, the annotation criteria may vary from time to time and from person to person. The annotators have to make continuous efforts to keep the annotation data consistent. In this paper, we propose a new annotation transformation method, called PATS, based on the patterns automatically detected from the source and target annotation data. The transformation patterns can process annotation items by means of structure mapping as well as flow control such as iteration. The PATS can also be applied to knowledge exchange, in which the concepts, attributes and relations of annotation resources are mapped into a certain specific format such as OWL. As a preliminary experiment, a Chinese poetry annotation project is used to test the effectiveness of PATS. The transformation result is represented in standard OWL format automatically and can be operated by Protégé directly. The experiment proves the effectiveness and efficiency of the pattern-based annotation transformation schema.