Hypermedia – Trellis Hypertext Reference Model
2.6 Trellis Hypertext Reference Model
The Trellis Hypertext Reference Model or “r-model” looks at hypertext as different levels of abstraction [Furuta & Stotts, 1990]. In general, hypertext can be divided into:
- Abstract Level: This layer is made of abstractly defined independent components that are connected together in some fashion. It does not describe the details of presentation.
- Concrete Level: Concrete representations in which the characteristics of the hypertext’s physical display have been established. That is, the contents of each of the windows is specified but not laid out.
- Visible Level: This layer is responsible for the layout and presentation of the hypertext network on a physical display.
The representations in the abstract level are at the greatest level of abstraction while those in the visible level are the lowest level. Whereas the abstract level can be standardized using document exchange protocols such as SGML, the visible level can be standardized using X Windows.
While some researchers believe that hypertext can be standardized, there are others who contend that the current notion of hypertext is not mature enough to be standardized. There is the third group which believes that currently identified and well-established features of hypertext may be standardized [ECHT '90 Panel Discussion, 1990]. Hardt-Kornacki et al., feel that “the components of hypermedia that are ready for standardization are not necessarily hypermedia-specific and the hypermedia aspects of these systems are not yet ready for standardization”. They contend that objects which have data and structure need rich and flexible standard representations in matters of exchange and authoring. However, since the same objects might be involved in non-hypermedia applications, they question the prudence of hypermedia-based object presentation standards. However, it is not clear if they mean that models such as Dexter and Trellis are not yet required. Also, standardizing the user interface for hypermedia systems is premature and naive. They add that it may be beneficial to standardize a generic set of tools that can be used to build various components of hypertext. They contend that none of the supporting infrastructure in a hypermedia environment such as display devices, user interfaces, file systems, broadband communications etc., are standardized.
However, such layered approaches (back-end, engine, and front-end) require significant efforts from software developers in writing exchange protocols between the layers. If these protocols are application dependent, they cannot serve as standard approaches. Hence, Isakowitz feels that alternatives to the layered architecture approach should also be explored [Isakowitz, 1993]. Mechanisms have to be devised to automate this layer-to-layer communication so that they are independent of applications.