As all the definitions described in the previous section can fit together, the recognition of such a diversity of conceptualizations, and the taxonomization of all of them, could be the basis of a neutral conceptual model. This new model would eventually provide the users with a number of different functionalities, adapted to each particular concept of learning object, and not necessarily restricted by only one of them. The flexibility is dealt with as a key issue for guiding the implementation. Based on a sound semantic model, defined by an ontology schema engineered using the Web Ontology Language (OWL), such a model would include all the definitions in McGreal’s study, thus supporting different types of learning objects. In short, the repository clients –final users, agents, and learning management systems– could, among other functionalities, add, retrieve, modify and search for learning objects regardless the definition of learning object used by the learning object creator. For example, client software using a LOM-based model would be able to retrieve learning objects from such repositories, even though the objects that fulfill the requirements of the client system were stored by a different system that was using the SCORM definition of learning object. From the different characterizations of learning object by McGreal, an ontology has been created (SLOR owl-file).

-- figure 2--

If learning can be considered an Event, any term linked to the representation of learning activities, or declared to have an educational purpose in some way, may be considered a learning object for practical purposes. In Figure 2, all the classes pre-fixed by “oc” are classes in the original Opencyc knowledge base. Therefore, the term learning is represented by the class oc_Learning, which represents the definition of learning in OpenCyc. The abstract representation from which all the terms in McGreal’s terminology derive is defined as the LearningObject-Generic class. An instance of this class may be anything used_in oc_Learning. The LearningObject-AsAnything class encompasses the broad definition of every possible meanings of learning object. The LearningObject-AsAnythingWithEducationalPurpose class has been defined to represent objects for which some declaration of their pedagogical purpose exists, while the class LearningObject-AsAnythingDigital represents digital objects. These two latter characterizations are combined in current specifications of learning technology. In addition, as current learning object standards and specifications suggest, an individual of the class LearningObject-AsAnythingDigital should be linked to at least one LearningMetadataRecord, by stating that the domain of the property hasAssociatedMetadataRecord is the class LearningMetadataRecord. LearningMetadataRecord is a generic term that can be used to derive specific terms supporting each particular specification such as LOM_Record or SCORM_SCO_Manifest. Any specification-specific learning object is, by its own nature, something with a declared educational purpose, which is at least tacit in the standardized schema. Figure 2 shows the relationships between all the terms, as described.
In this figure, classes in the ontology are shown as solid-outline rectangles, with several compartments separated by horizontal lines. The name of the class is held in the top compartment. The rest of the compartments hold both the properties of the class, and the restrictions defined for it. Object properties are shown preceded by a circled “o”, while restrictions are preceded by a circled “R”. For example, the class LearningObject-Generic is linked to the class oc_Learning by an arrow that represents the object property used_in in the corresponding compartment of the class. Note how the arrow is labeled with the name of the property, used_in in this case. The figure also shows the hierarchical relationship between the classes in the ontology. Classes representing more generic concepts appear in the top part of the figure. The hierarchical relationship is shown as a solid path from the more specific classes (e.g. LOM_Record) to the more generic ones (e.g. LearningMetadataRecord), with a hollow triangle at the end of the path where it meets the generic class. Other links, like those labeled with the intersection and the equivalence symbols, represent the kind of formal definition for a class as stated in the ontology. The class LearningObject-AsAnythingDigital, for example, is linked to an intersection symbol by an arrow labeled with an equivalence symbol. This is to represent the fact that the class LearningObject-AsAnythingDigital is defined as the class of objects that are instances of two classes simultaneously. In this particular example, instances of LearningObject-AsAnythingDigital accomplish the properties in both the class oc_Learning and the class LearningObjectAsAnything.

Some examples:

LearningObject-AsAnyThing: Books, notes, applets, web pages, a paper with schemas, questionnaires, etc.
LearningObject-AsAnyThingDigital: web pages, applets, digital papers, programs, e-books, etc.
LearningObject-AsAnyThingWithEducationalPurpose: books with educational purpose, practices, exercises, questionnaires, etc.
LearningObject-LOM or LearningObject-SCORM_SCO: digital lessons with educational purpose – Cisco Course, Moodle Course, ATutor Course, etc.