next up previous
Next: Computing Without Syntax! Up: Semantic Computing Vision of Previous: Implementation

Possibility of Semantic Web Without Unicode Markup

A notable feature of GNOWSYS is the provision of a unique URL for each object in the knowledge base. This feature is essential for using GNOWSYS as a distributed publishing tool for large structured knowledge bases like thesauri, concept bases, encyclopedia, and any such information systems including making regular dynamic web sites. But, most enabling aspect of this feature is for developing semantic web. GNOWSYS provides a unique opportunity for developing semantic web without regular use of markup. This is because GNOWSYS is a server listening to structured queries directly over the Internet. In the current practice the database schemes were hidden due to the intervening markup layer. While XML solves this problem by declaring the structure, it does it in an expensive way. Let us see how.

A client to server communication is an agent to agent communication. No human being is required to make the client understand the markup. Therefore, what we need is a binary markup between artificial agents, and a translation between binary markup and UI for us to see the structure and semantics. Currently the documents in WWW are structured using a markup mostly in ASCII or Unicode. It made good sense because the communication was mainly intended to be between a machine and human being. In future when the web extends to semantic web, a lot of communication is expected to take place between machines. Why should machines need such expressive (read expensive) markup, if human beings are not reading the structure?

GNOWSYS model, when frozen, like all databases will have a finite number of knowledge organizers for structuring all structurable data. Unstructured data will of course pass as Unicode. Such structured data can go without markup in Unicode, but a more economical code, reducing the required data transfer and also the need to parse. Unicode markup is possibly required for programmers to give structure, and for data exchange between heterogeneous systems. The very idea of Unicode based markup like XML was invented to meet this requirement. But, since we are slowly moving towards standards for knowledge exchange, we should move towards databases structured with a standard structure. Databases can be heterogeneous in their implementation strategies, but could nevertheless agree on standard knowledge exchange schemes, just as computing industry adopted Unicode for human interphase. If, e.g., OWL becomes a standard, then databases should be made for storing OWL schema, rather than working with fat XML files. GNOWSYS or such systems can help us move in this direction.

When this happens a browser can directly render the structured knowledge directly from the knowledge base into either textual form, controlled natural language, tabular form, concept graphs or any such human interphase which preserves the structure and semantics. An advantage of rendering structure on the client side is the user can choose the preferred mode of drawing depending on the toolkit the client uses. Additional advantage is eliminating (1) the creation or storing of structure as Unicode based markup on the server side and (2) interpreting such markup on the client side.

Further, GNOWSYS uses a GQL, Gnowledge Query Library, not a language. Since this is a library, the query and management functions can be directly embedded in any programming language, eliminating the need to perform additional parsing on the server side. Agent oriented computing becomes easier since the agents talk directly to the knowledge base using GQL.

Computers had demonstrated their potential that they can `read' and `reply' more efficiently as databases, and programmers had demonstrated that they can manage databases easily. Massive deployment of databases everywhere and by anybody, is a good indication that this approach works. By transforming web servers into knowledge bases with RPC support we can publish large scalable semantic web sites, and not by serving through web servers publishing fat XML files containing knowledge bases.


next up previous
Next: Computing Without Syntax! Up: Semantic Computing Vision of Previous: Implementation
Nagarjuna G. 2005-01-25