Metadata and Metamodeling


Schema-based Structured Models

Another very common method of modeling metadata is to use markup definition languages to define a structural metadata model. In this method, Document Type Definitions (DTD) or XML schema languages (i.e. XML Schema [116], Schema for Object-oriented XML (SOX) [33], or RELAX NG [109]) define a tree structure of XML tags. Each element represents a piece of metadata element, and the hierarchy represents relationships such as parent-child, container-subelement, and group members. Depending on the schema, metadata could be extendible as well as very strict leaving no space for change in structure.

In schema-based metadata solutions, structure of metadata is more important than the meaning of metadata itself, which:
  • allows traditional data models, i.e. E-R diagram or object-oriented design models, to be easily represented in XML,
  • simplifies mapping persistent data to in-memory objects and vice versa (as in data binding implementations [51, 103]), 
  • allows already existing XML tools, storage, management, and query systems to be used for schema-based metadata without compatibility problems raised.
Figure 2.11 depicts the same object model represented in UML [Fig. 2.11(a)] and XML [Fig. 2.11(b)]. Person is the parent class of Employee. In the example, XML Schema language is used to model the XML data that can be instantiated as in figure 2.11(d). UML instantiation is shown in Java language in figure 2.11(c). Structural similarities between the models allow system developers to transform one model into another very easily.

The number of possible models that structural modeling languages can define is infinitely large. This gives organizations freedom to define proprietary models at will. However, such freedom causes many duplicate models to be created. To solve this problem, national and international standardization organizations and working groups take initiatives to organize such efforts.

Among these efforts, IMS Project’s LIP (Learner Information Package) [64], ADL’s SCORM (Sharable Content Object Reference Model) [2], and IEEE’s LOM (Learning Object Metadata) [75] are worth mentioning. All of these initiatives that model metadata for learning resources and activities started independently, and, at the time, are very much integrated to offer a complete set of standards. Learning resources are among components, such as learners (i.e. students, remote on-line users) and teaching materials and curricula, that form a learning management system (LMS). Content developers can follow specifications to simplify the process of describing learning resources, and more importantly, to ease the exchange of metadata about content and learners between LMSs.