CREAM: CREAting Metadata for the Semantic Web

Abstract

Richly interlinked, machine-understandable data constitute the basis for the Semantic Web. We provide a framework, CREAM, that allows for creation of metadata. While the annotation mode of CREAM allows creation of metadata for existing Web pages, the authoring mode lets authors create metadata––almost for free––while putting together the content of a page.

As a feature of our framework, CREAM allows creating relational metadata, i.e., metadata that instantiate interrelated definitions of classes in a domain ontology rather than a comparatively rigid template-like schema such as Dublin Core. We discuss some of the requirements one has to meet when developing such an ontology-based framework, e.g., the integration of a metadata crawler, inference services, document management and a meta-ontology, and describe its implementation, viz. OntoMat, a component-based, ontology-driven Web-page authoring and annotation tool.

Introduction

The Semantic Web builds on metadata describing the contents of Web pages. In particular, the Semantic Web requires relational metadata, i.e., metadata that describe how resource descriptions instantiate class definitions and how they are semantically interlinked by properties. We have carried through several case studies that build on this idea of the Semantic Web in order to provide intelligent applications to make knowledge about researchers, about companies and markets, and about research papers accessible by semantic means. An important cornerstone of the case studies––and many other scenarios in the Semantic Web––is a framework and a mechanism that let users easily and comprehensively contribute relational metadata that the Semantic Web builds upon. For this objective, we have developed our semantic annotation framework, CREAM––CREAting Metadata for the Semantic Web––that we present here.¹ CREAM is geared to allow for the easy and comfortable creation of semantic metadata.

CREAM allows for the a posteriori annotation of existing resources. A posteriori creation of metadata involves the consideration of an item (e.g., a document) by an agent (possibly, but not necessarily, a human agent) and its description with metadata. For this purpose we distinguish two modes of interaction: (i) the process may include the identification of elements already existing within the item. For instance, one may consider an HTML page and identify its author by a footer appearing within the page; (ii) one may classify the item as belonging to a category like Business though neither the term itself nor a synonym or hyponym appears in the item.

A posteriori creation of metadata comes with a major drawback. In order to provide metadata about the contents of a Web page, the author must first create the content and second annotate the content in the additional, a-posteriori, annotation step. To avoid the work overhead, we propose a third mode of interaction, (iii) an author has the possibility to easily combine authoring of a Web page and the creation of relational metadata describing its content.

Scrutinizing the differences between the modes of interaction (i), (ii), and (iii), we found that the major problems one must deal with for an annotation framework are identical. In fact, we found it preferable to hide the border between annotation (i.e., (i) and (ii)) and authoring (i.e., (iii)) as far as possible within CREAM. Some questions, however, were triggered by the need to distinguish the ontology that defines the structure of the relational metadata from its use for annotation or authoring. For this purpose, we introduce here a meta-ontology that describes how the annotation and authoring modes of OntoMat interfere with classes and properties of the ontology proper. We modularize the ontology parts needed in the metadata creation process from the ones relevant for the targeted content description.

For the CREAM framework we have also provided a reference implementation, viz. OntoMat-Annotizer (OntoMat for short), which is freely available for download.²

In the following we first describe the case studies from which we took a major part of our experiences for guiding the development of CREAM (Section 2). Then, we describe some of the requirements in detail that were derived from the case studies (Section 3). We explain our terminology in more detail and give an example of the metadata we want to create in Section 4. We derive in Section 5 the design of CREAM from the requirements elaborated before. In Section 6, we specify how the meta ontology may modularize the ontology description from the way the ontology is used in CREAM. In Section 7, we explain the major modes of interaction with OntoMat, our implementation of CREAM. Before we conclude, we give a survey of related work in the areas knowledge markup on the Web, knowledge acquisition, annotation environments and authoring environments.