INTEROPERABILITY OF LEARNING OBJECT REPOSITORIES: COMPLICATIONS AND GUIDELINES

ABSTRACT In this paper, we present an approach for interoperable metadata, using application profiles. As a case study, we map the instances of the ARIADNE application profile into instances that conform to the IEEE Learning Object Metadata (LOM) standard. We use XSLT to transform ARIADNE XML instances into IEEE LOM Instances. The major contribution of this paper is that it explains how mapping and transforming metadata between different metadata specifications is not always a simple one-to-one process. We present the different classifications of metadata application profiles and discuss issues related to developing such profiles. Also we present and analyze the different kinds of complications involved in mapping profiles into standards. KEYWORDS Interoperability, Learning Object Repositories, Me tadata, Metadata Mappings, Application profiles 1. INTRODUCTION More and more systems rely on some metadata standard for representing metadata [Duval, E., 2004 & Rehak, D., 2003]. Standards facilitate sharing and exchange of learning objects as well as their metadata. Metadata is defined in this context as “information about an object; be it physical or digital” [Duval, E., 2001]. It is used to facilitate search, evaluation, acquisition and use of learning objects. Typically, Learning objects and their associated metadata are located in distributed Learning Object Repositories (LOR’s). Different Learning Object Repositories try to address different needs. Therefore, metadata designers may select specific metadata elements, as well as the value sets they derive their values from, from one or more metadata standards [Heery, R., 2000]. The specification of these metadata elements and value sets is called an “application profile”. Profiles are used to adapt metadata specifications to the requirements of the local community, taking for instance multilingual and multicultural requirements into account [Duval, et al., 2002]. Examples of profiles are SingCore [SingCore, 2002], SCORM [ADL, 2001] and ARIADNE [ARIADNE, 2001]. In order to represent in a standardized way the syntax and semantics of metadata elements in an application profile, as well as their value sets, we need to conceptually map the data elements and value sets of a profile into elements and value sets of a standard schema. Typically, we rely on a representation (technical) such as XML or RDF for this purpose [Euzenat, J., 2001]. In this paper, we map the ARIADNE profile into the LOM standard. We use XSLT to transform ARIADNE XML into conforming IEEE LOM XML. More generally, we provide guidelines for conceptually developing and mapping metadata of profiles and discuss different important issues related to technically transforming and validating metadata. This will help other Learning Object Repositories to map, transform and validate their metadata. The paper is structured as follows: In section 2, we present the different types of application profiles, and discuses issues related to the development of such profiles. In section 3, we discuss the different kinds of metadata mappings used to map profiles into metadata standards. Section 4, presents the mapping of profile metadata elements into elements of standards. Section 5, presents the mapping of profile value sets into value sets of standards. Finally, a conclusion is given in section 6.