Site Map
Schematic
Concept of Operation
Discussion
Home

Vision - the Dynamic Web

  • My vision is that the web should grow into a giant knowledge base - a global repository of facts and perceptions that can be readily incorported into any number of documents.
  • The aim here is to get right down to the underlying knowledge that is presented in documents so that it can be shared between sites and updated automatically in response to fresh intelligence.

A Web of Data about Everything for Everyone

  • The World Wide Web Consortium has an aspiration that the present and highly successful "web of documents" should grow into a "web of data about everything for everyone". To my mind this means that anyone who has information they would like to share must be able to add that information to a single coherent global knowledge base.
  • I call this the dynamic web because the world around us is changing all the time. A global "web of knowledge" must cope with change and competing versions as a matter of routine. To achieve this we need a form of publish and subscribe service to distribute fresh intelligence to the various sites at which the affected documents are held.
  • Through this global knowledge base the ordinary user would be empowered to share knowledge with other ordinary users without recourse to programming skills.

Keeping it Simple

  • My idea is that we should develop standards that allow the rules for updating and distribution of dynamic information to be specified entirely in the HTML of the pages concerned. Information specified in accordance with these standards would be held in a site knowledge store that forms part of a global knowledge base available to all comers.
  • The key that makes this technically simple is to ensure a consistent granularity for all information whatever its source. This is achieved by insisting that all information on the dynamic web is broken down into grains of knowledge represented by assertions that can be regarded as "indivisible and immutable versions which are semantically complete and have their own provenance".

Assertions

  • All information on the dynamic web has the same format which is based on the following five components:
  • Subject. This is a URI that says what the information describes. This URI may identify either:a real-world object or a part of the global ontology, i.e. a class, class specializing scheme or property.
  • Predicate (meaning) . This is a URI that identifies a property defined within the dynamic web as part of a global ontology. This property provides the complete and only explanation of how this citing of the object (the value) is to be interpreted.
  • Object (value). This is the information itself - expressed either as a URI (i.e reference to some other resource) or as text tagged in accordance with an XML Schema defined for the Property that is cited in the predicate.
  • Context. This is a URI that identifies a set of constraints within which the assertion is valid. These constraints may include geographic, administrative, time and other bounds.
  • Provenance. This is needed so that subscribers can apply their own criteria to determine the relevance of each assertion to their own needs. It comprises the following detail in the form of XML tagged text:
    • The URI of a person or organization that is the immediate source.
    • The time-stamp of posting onto the dynamic web.
    • The URI of the site at which the assertion was first posted onto the dynamic web.
    • The time period over which the information is deemed to be valid.
    • Further optional detail such as:
      • the URI of the ultimate source of the information.
      • the URI of a recognized definition of confidence level.

A Global Knowledge Base

  • Initially, each of these site-based pools of dynamic knowledge data would be visible only through pages on the same site, but they merge into a global resource as and when individual sites choose to co-operate through the following simple procedure:
    • A willingness to share some of this knowledge data with other sites is signalled on public pages with an offer to publish.
    • Any site for which this offer is accepted becomes a subscribing site that will receive copies of each new version of the knowledge data concerned.
    • These copies are added to the subscribing site's own site knowledge store as and when received.
  • Each participating site thus includes a highly selective subset of the global knowledge base with content chosen and controlled by its own webmaster. Any information from this subset can then be incorporated into any number of individual pages within that site. This will make it easy to create pages that collate knowledge data from a range of sources with an assurance that this information is kept up-to-date.
  • The page called "The 'Place of Choice' for All Knowledge" offers an engineer's view that attempts to show in outline how the vision of a dynamic web builds on existing W3C recommendations and current practice to enable a "Web of data about everything for everyone".

A Single Global Ontology

  • Semantics are also crucial to the fulfilment of this vision. If knowledge is to be freely combined, the meaning cited for each assertion must come from a single ontology. To work on a global scale this ontology must cover every way in which things can be classified and every type of information through which knowledge can be expressed.
  • It may seem that an ontology with this scope would be so vast that it would be impossible to create, agree or maintain. But what if the definitions of classes and properties were themselves regarded as part of the global knowledge base? The ontology itself would grow from individual initiatives and be distributed in exactly the same way as other information. A global ontology like this would have to allow for each class to be specialized in any number of ways - but this is not difficult to do! Some arguments in support of this claim are to be found in white paper entitled "A Global Ontology for a Truly Dynamic Web" that was given at the Second Semantic Technology Conference, San Jose, March 6-9, 2006.
  • The key to this global ontology is that each class may be specialized in any number of ways through the creation of distinct class specializing schemes. Once a class is published onto the global ontology, it may be used as a root class for many different schemes that meet the diverse needs of other participants in the dynamic web. The member classes of all these schemes inherit the definition of the chosen root class but they do not constrain that class in any way.

A Plan of Action

  • I wish to show that the software needed to support the dynamic web is very simple and to contrast this with the immense potential that it offers. The functions are simple and independent of the information handled. This means that there is no significant investment threshold to deter people from joining in. The vision can begin to materialize through numerous unconnected initiatives - each involving just a handful of collaborators.
  • The beauty of this concept is revealed as these initiatives begin to overlap. The many fragments of information will all fit together so that they grow into a single global knowledge base without conflict while preserving the many varied perceptions from it is fed. There is no need for any major commitment from anyone. All we need to do is create the necessary standards and the enthusiasm and immagination of ordinary Web users will do the rest.
  • I intend to set up a demonstration to show how the envisioned standard for assertions combines with that for class specializing schemes to enable a truly global web of knowledge data. The idea is that this can be based entirely upon simple freeware, written by myself if necessary, with support from readily available implementations of existing W3C trcommendations. Work on this pilot implementation is progressing well such that all the main facilities are now in place. I will make this software available to interested persons just as soon as I feel it to be sufficiently robust.
  • It is my hope that some like minded persons may share this vision and be inspired to collaborate or assist in some way - please contact me at harryellis@acm.org.