RDFa Based Annotation of Web Pages through Keyphrases Extraction

The goal of the Semantic Web is the creation of a linked mesh of information that is easily processable by machines, on a global scale. The process of upgrading current Web pages to machine-understandable units of information relies on semantic annotation. A typical process of semantic annotation includes three main tasks: (i) the identification of an ontology describing the domain of interest, (ii) the discovering of the concepts of the ontology in the target Web pages, and (iii) the annotations of each page with links to Web resources describing the content of the page. The goal is to support an ontology-aware agent in the interpretation of target documents. In this paper, we present an approach to the automatic annotation of Web pages. Exploiting a data reverse engineering technique, our approach is capable of: recognizing entities in Web pages, extracting keyphrases from them, and annotating such pages with RDFa tags that map discovered entities to Linked data repositories matching the extracted keyphrases. We have implemented the approach and evaluated its accuracy of on real Web sites for e-commerce.