## Semantic Transcoding: What we should do now for building the Semantic Web

Katashi NAGAO
Center for Information Media Studies, Nagoya University

### 1 Introduction

The Semantic Web as a next generation information infrastructure for knowledge sharing requires some tools and platforms to create and distribute semantically extended Web content.We have been developing such system that has almost same functions as theSemantic Web along the straight line of an extension of the current Web.We call our framework Semantic Transcoding'' because one of our main purposesis to trancode (personalize or adapt) Web content into more appropriatepresentation format using semantic information about the content.This framework consists of the following three subsystems:

1. An authoring tool to annotate existing Web content with additional information that can facilitate machine's understanding of the content

2. An annotation server that manages annotation data to share the data

3. A transcoding proxy that transforms and adapts content to users on demandusing its annotation data

Based on our experiences on development of these systems, I would liketo recommend what and how we should do for building an instance of the Semantic Web as early as possible.

### 2 What to do for building the Semantic Web

The Semantic Web aims at a global knowledge sharing infrastructure. However, it is too challenging to realize itswhole functionality along with the plan.Artificial intelligence and knowledge management researchers have beentried to create a general methodology for universal knowledge representationbut failed.How we

I guess that there are some critical issues of the current (planned)'' Semantic Web.The examples are as follows:

1. Language specifications of the components of the Semantic Web such as RDF (Resource Description Framework) and OWL (Web Ontology Language) are too complicated to learn.

2. Literacy building for metadata authoring is not yet establishedand ordinary users are not well motivated to create metadata.

3. No concrete applications are proposed in detail except semantic retrievalof content.

4. Semantic descriptions of non-textual content such as multimedia data are not proposed in the context of the Semantic Web.

Below I discuss some practical solutions for the above issues.Key concepts include a hierarchization of information based on annotations and an individualization of information via transcoding.We employ XML (eXtensible Markup Language) as a basic framework of interoperability of structured information.

An extension of present Web that is realized by annotation and transcodingin terms of its content and usage is also a realization of the SemanticWeb

### 3 Annotation: Hierarchization of Digital Content

Fugure1: Extension of Web Content by Annotations

### 5 Mechanism of Transcoding

Fugure2: Overall Configuration of Semantic Transcoding System