Ontology-Based Description of Learning Object 469 many researchers focus on the description about LO. There are also many standards about learning object metadata; however, the metadata is not enough in fact. The metadata just provides some descriptions of LO’s properties, it is lack of reasoning capability and reusing capability, we will discuss in detail in Part 3. In this paper, the authors pay more attention to the ontology driven description of LO. At last this paper presents the approach to build learning object ontology. The benefits of such approach are increasing the share and reuse abilities of Learning Object. 2 Learning Object Educational content on the Web is often called Learning Object terminological today. Although the Learning Object is accepted widely, the discussion about its definition still remains. So far there isn’t a generally accepted definition within the learning object community. 2.1 Definitions and Characters of Learning Object What are learning objects? There are three prominent characterizations[2]: "Modular digital resources, uniquely identified and meta-tagged, that can be used to support learning" (National Learning Infrastructure Initiative); "Any digital resource that can be reused to support learning"(David A. Wiley); "Any entity, digital or non-digital, that may be used for learning, education or training"(LTSC, IEEE Learning Technology Standards Committee). All these definitions seem to be different formally but they describe the inherent characters of learning object. 2.1.1 Share and Reuse All of the researches regard reusability as the core of learning object. Other characters are set to serve reusability. From the standpoint of learning object reusability, it would be advantageous for learning object to have many different uses, so that expensive multimedia content elements could be reused in as many different learning objects as possible[3]. Learning object can’t be reused widely unless it can be shared. "Share" is a key- word and a core problem when we use the digital resources today. Share enables peo- ple to use the same resource at different time, in different places and for different goals. 2.1.2 Digital The digital processing is inundante in information time. "Digital" separates the learning object from the traditional education resources. LTSC defined learning object is "digital or non-digital" entity. The point is all the resources can be managed digital. 2.1.3 Metadata-Tagged Each learning object contents a series of description information that is metadata (like title, author, version, format, content description and instructional function). The metadata is indispensable that ensures learning object can be retrieval easily and 470 X. Wang, F. Fang, and L. Fan effectively; and ensures to transport and exchange learning object between different applications. 2.1.4 Instructional and Target-Oriented Learning object has a clear target. It generated for instruction and existed for instructional design. Each learning object should have a definite target. The target is for learners in different level to know, master and apply some learning content. After using this learning object the learner should achieve the relevant instructional target. 2.2 Learning Object Metadata In order to describe, organize, access and reuse learning object efficiently, a lot of researchers and instructors devoted themselves to the development of learning object metadata. Why the metadata is needed? The Learning Object is lack of integrality without metadata. Metadata provides better representation and understanding of learning con- tent, and enables people to transform, share and reuse learning content. Learning object metadata fulfill many roles. One of them is the role of an indexing system that allows instructors and authors to find educational content easily that matches their instructional needs. A high-level structure of learning object should contain both its content and metadata. So far several metadata standards and specifies have been developed aiming at im- proving the reusability. For example, IEEE Learning Object Metadata (IEEE LOM, 2002) and Dublin Core are two important standards abroad which specify a standard- ized set of metadata that facilitates retrieval of Web-based resources. 2.2.1 LOM The typical metadata standard for learning object is the Learning Object Metadata (LOM) standard, defined by the IEEE Learning Technology Standards Committee. It defines fields for describing learning object in terms of their general properties (such as title, size, language and author), technical requirements (like type, format and dura- tion), pedagogical characteristics (like context difficulty, interactivity type, etc), and so forth. LOM defines basic schema that is a hierarchy of data elements for learning objects metadata. There are nine categories (General, Lifecycle, Meta-metadata, Technical, Educational, Rights, Relation, Annotation and Classification) at the top level of the hierarchy in LOM[4]. For each data element, LOM specifies a name, explanation, size, example value, data type, and other key details. 2.2.2 CELTS The CELTS[5] is a series of standards and specifications established by China E- Learning Technology Standardization Committee. They references many famous standards like IEEE LOM, DCMES, the taxonomy from GEM and EdNA, and some standards or drafts based on the education status quo in China. CELTS-42 Metadata Application Specification of Basic Education Resource. It is the main reference for our processing. Ontology-Based Description of Learning Object 471 2.3 The Metadata Is Not Enough Although the definition of learning object seems clear and the metadata is mostly specified but the metadata is not enough. The metadata just lists some of the proper- ties about LO, thus, many questions still to be answered, such as: from the standpoint of learning object reuse, it is content reuse, how can we describe the content of LO formally? Can LO has stronger reasoning ability? How can we reuse LO in different learning context? To address these questions, we introduce an ontological methodol- ogy to describe learning object. 3 Ontology Ontology is the core of Semantic Web technology, and in recent years the develop- ment of ontology has been moved to the desktops of domain experts. Ontology has become common on the World Wide Web, and now let’s start from what ontology is. 3.1 Definitions "Ontology" has a long history in philosophy which means the subject of existence. But this term we use here refers to "explicit formal specifications of the terms in the domain and relations among them" (Tom Gruber, 1993). Thus ontology is an explicit formal specification of the terms in the domain and relations among them [6]. Ontol- ogy defines a common vocabulary for researchers who need to share information in the domain. It includes machine-interpretable definitions of basic concepts in the domain and relations among these concepts. In this research we use Gruber’s definition and in which specification means a formal and explicit representation. In a good structural ontology all the concepts are described by formal language and this formal representation makes the ontology ma- chine-readable. The concepts, classes and relations between them are well defined in ontology. The relations among different concepts are specifies which enable different programs to use the same knowledge. Conceptualization is another character of ontology which refers to the abstract and simplified view of the world [1]. The "world" can be some substances or some phenome- non. In ontology’s world everything is represented by abstract concept, i.e. any formally ontology is based on a certain conceptualization. Every conceptualization is based on the concepts, objects, and other entities that are assumed to exist in some area of interest and the relationships that hold among them. Why would someone want to develop ontology? To answer this question let’s list the advantages of Ontology. Ontology enables people or software to share common understanding of the information structure; enables people to reuse domain knowl- edge and to make domain assumptions explicit; helps people to analyze domain knowledge and separate domain knowledge from the operational knowledge [6]. It is clear that ontology is about domain knowledge and the core of ontology is sharing and reusing. 472 X. Wang, F. Fang, and L. Fan 3.2 Ontology and Education Applying ontology to education is a new research field, and researchers have devel- oped some educational Ontologies abroad (such as OntoEdu). In the beginning of WBE people worried about that the resources are not enough so lots of digital re- sources are uploaded to the Internet and plenty of repositories emerge endlessly. It’s not exaggerative to say that nowadays resources are abundant or inundante and any resources can be digital nearly, meanwhile the quantity is not a problem at all. But new challenges appear with the increase of resources: How to store and organize resources efficiently and how to enhance the capability of sharing and reusing? How can learners search and retrieval the resources they need as quickly and accurately as possible? How to ensure the learners can achieve the learning goal with the help of this resource? The development of Semantic Web and Ontology brings fresh air to WBE. Ontol- ogy represents domain knowledge by defining terminology, concepts, relations, and hierarchies. They enable different education applications to share and reuse the same educational content. Furthermore, Ontology is machine-readable and it can provide semantic foundation to achieve sharing overall. It will be faster and more convenient to query and retrieval educational material with the Semantic Web services. 4 Ontology-Based Description of Learning Object 4.1 Ontology Is Necessary The learning object metadata is specified, but it is lack of reasoning ability and the machine processing ability if using metadata only. Note that ontology is good at rea- soning and machine-readable as expatiate in Part 3. The benefits of ontological repre- sentation of domain knowledge lie in its capabilities of explicitly defining concepts and their attributes and relationships. Coupled with new information technologies, such representation can be encoded in ways that allow for direct conversion into im- plementation models. Ontological modeling for learning objects may be divided into three broad areas: content, presentation, and application [7]. As a conceptual model, Ontology can express the basic knowledge system of a domain. Basic on the research of the necessity and feasibility to a apply ontology to Web resources, we proposed to depict the learning object semantics by ontology and implemented in the Semantic Web framework, thereby improving the learning object sharing through semantic interoperability. 4.2 Learning Object Ontology (OntoLo) A Learning Object Ontology for basic education was created here. An effective tool is necessary to develop ontology. We investigate a variety of ontology developing tools (OilEd, OntoEdit, WebOnto, OntoSaurus and Ontolingua) and select Protégé (http://protege.stanford.edu ) to establish this ontology. Then we present some formal descriptions of this ontology here. Ontology-Based Description of Learning Object 473 Fig. 1. A Protégé screenshot representing the class hierarchy of OntoLo 4.2.1 Classes There are several possible approaches in developing a class hierarchy, like top-down, bottom-up and combination development processes. In this study the top-down devel- opment process is much more suitable. But the problem is the taxonomy of learning object is not uniform. David Wiley said "All learning objects have certain qualities. It is the difference in the degree to which they exhibit these qualities that makes one type of learning object different from another"[7].According to the situation of the primary education in China mainland we created three basic classed here: Class: Learning Objects, Class: Learning Objectives and Class: Categories. A Protégé screenshot represents the classes hierarchy of OntoLo show in Fig.1. 4.2.1.1 Class: Learning Objects. Learning Object is used to describe the subjects in primary education domain. Considering the expansibility all of the subclasses are defined based on the Subject Classification And Code from Ministry of Education. 4.2.1.2 Class: Learning Objectives. This class describes the different target levels of learning. Learning objectives are composed of Action, Competency and Context[7]. There are 23 sub-classes of this class in total, and we can update these objectives when needed. . share information in the domain. It includes machine-interpretable definitions of basic concepts in the domain and relations among these concepts. In this research we use Gruber’s definition. standard for learning object is the Learning Object Metadata (LOM) standard, defined by the IEEE Learning Technology Standards Committee. It defines fields for describing learning object in terms of. terminological today. Although the Learning Object is accepted widely, the discussion about its definition still remains. So far there isn’t a generally accepted definition within the learning