Semantic Annotation and Transcoding: Making Text and Multimedia Contents More Usable on the Web

Katashi NAGAO
Dept. of Information Engineering, School of Engineering, Nagoya University


This paper proposes an easy and simple method for constructinga super-structure on the Web which provides current Web contentswith new value and new means of use.The super-structure is based on external annotations to Web documents.We have developed a system for any user to annotate any element of any Webdocument with additional information.We have also developed a proxy that transcodes requested contents byconsidering annotations assigned to them.In this paper, we classify annotations into three categories.One is linguistic annotation which helps the transcoder understandthe semantic structure of textual elements.The second is commentary annotation which helps the transcoder manipulatenon-textual elements such as images and sounds.The third is multimedia annotation, which is a combination of the abovetwo types.All types of annotation are described using XML, and correspondencebetween annotations and document elements is defined using URLs and XPaths.We call the entire process ``semantic transcoding'' because we dealwith the deep semantic content of documents with annotations.The current semantic transcoding process mainly handles text and videosummarization, language translation, and speech synthesis of documentsincluding images.Another use of annotation is for knowledge discovery from contents.Using this idea, we have also developed a system which discovers knowledgefrom Web documents, and generates a document which includes the discoveredknowledge and summaries of multiple documents related to the same topic.