Weblog Kitchen
Welcome Visitors
Things To Do


Recent Changes
Most-linked pages
Whos Who

Hypertext Kitchen

Style Guide
How To Edit
Wiki Sandbox

In 1990, researchers at the UniversityOfSouthampton identified four problems common to hypertext systems at the time:

  1. Authoring effort required Documents become available faster than they can be converted for a hypertext system.

  2. Closed systems Hypertext systems usually run as stand-alone systems that will not communicate with other software packages. This made it difficult to extend systems and cope with new data formats.

  3. Proprietary document formats It is not possible to take proprietary documents created in one system and use them in the hypertext system. This compounds the authoring effort problem.

  4. Read-only media Many hypertext systems used tags or pointers within the data to represent links. When that data is read only (e.g. a CD-ROM) this is not possible (unless of course the links are already in the read-only data, in which case they cannot be updated).

Microcosm was an open hypertext system designed to overcome these problems (in fact, these problems provide a common motivation for OpenHypertext research). Some important guiding principles for the design of Microcosm were:

  • No separation between author and user: all users can author links

  • Loosely coupled system: Microcosm is a set of communicating tasks, open in structure. Further components can easily be added to extend its functionality.

  • Separation of links from nodes

Microcosm allowed users to create a variety of different links:

  • Specific Links Link a specific anchor (e.g. text selection) in a source document to a specific anchor in a destination document.

  • Local Links Link any occurence of an anchor in a specific source document to a specific anchor in a destination document.

  • Generic Links Link any occurence of an anchor in any source document to a specific anchor in a destination document. Generic links greatly improved authoring efficiency (a link only need to be created once, and would even be added to new documents) and maintenance (only one link had to be changed if the destination changed). A trivial example of generic links is a dictionary lookup facility - every appearance of a dictionary term in any document would be linked to the definition of the term.

  • Dynamic Links Link specific anchor in a source document to a number of dynamically computed destinations (for example by using a text retrieval search).

  • Relevance Links Link to all other documents in the same cluster (documents are pre-indexed into clusters according to their relevance to each other).

Microcosm was intended to be used with any application. However, this could be a problem as some applications may not actually be "aware" of Microcosm. Applications were therefore classified into three types:

  1. Fully aware applications were written explicitly for Microcosm and could access full hypermedia functionality

  2. Partially aware applications were those which were able to be customised to receive some level of hypermedia functionality from Microcosm (for example, Microsoft Word could be made Microcosm "aware" through its macro language).

  3. Unware applications were those which could not be adapted. Documents belonging to these applications could act as the destination of a link (the document would simply be displayed by the application when the link was followed). To integrate unaware applications, Microcosm could monitor the Microsoft Windows clipboard and process data pasted there.

Links were created and followed in Microcosm using a process of selection and action. The user selected the information they were interested in (e.g. a word or phrase in a text document or an area in an image), and chose an action to be performed on the selection (e.g. make or follow a link). Messages to perform hypermedia actions were sent from the application to Microcosm and acted upon by a chain of filters. This filter model is a particular feature of the Microcosm architecture. Each filter manipulated messages by blocking, deleting, or adding to it. Messages emerged from filter chain to a link dispatcher process, which examined the messages and offered actions to the user (such as available links to follow). Microcosm filters included a number of linkbases, a history filter, and filters which computed dynamic links.

Navigation in Microcosm was also user-led: the reader selected the region of interest, and queried the system for links relating to the selection.

Users could create their own links, which could be stored in a private linkbase or shared within a user group. Users could also add and remove filters from the chain - a typical example is adding linkbases which reflect the context in which a particular set of documents is to be investigated (for example, selecting a linkbase which offers beginner, intermediate, or advanced dictionary lookup, depending on the user's experience of a topic area).

See Also:


Microcosm - An Open Model for Hypermedia With Dynamic Linking
Andrew Fountain, Wendy Hall, Ian Heath, & Hugh Davis, Proceedings of the European Conference on Hypertext 1990 (ECHT'90), 298-311, 1990.

Towards an Integrated Information Environment with Open Hypermedia Systems
Hugh Davis, Wendy Hall, Ian Heath, Gary Hill, & Robert Wilkins. Proceedings of the 1992 ACM Hypertext Conference, Milan, 181-190, 1992.

-- Last edited October 27, 2002

Weblog Kitchen | Contact

Sponsored by Eastgate