As a principal, when one takes to the task of solving a problem, it is of indispensable value to have ascertained a full grasp of that which is to be solved. He who endeavors upon such a problem lacking this will surely find a solution to nothing more than the problem of excess time. It is exactly this problem which I have lamentably solved time and time again for lack of the aforementioned principal.
I believe it was St. Thomas Aquinas who once observed "an error in the beginning is an error indeed." So as to dodge such folly as can be avoided I will be at some pains herein to set forth the problem, or problems as the case may be, which the intended system is meant to solve.
We start with the definition of a model which, in a sense, can be said to be the impetus for the problem itself, the Semantic Web. The notion and movement of the Semantic Web, as defined by those who define such things, is aimed at the bringing about of an interconnected web of data points, the connections between which are made meaningful by use of a common ontology. Work in this area is therefore performed in the space of resources and is concerned heavily with their creation, modification, destruction, and, most importantly, relation and aggregation. There are and have recently been made advances in systems for handling such operations. Many of these systems are satisfactory and some are quite mature. Thus it is not in this realm that I wish to focus my attention (though hitherto it has been this area to which my efforts have been employed).
As is, the Semantic Web would be a great boon for machines duly coded to handle the interpretation and aggregation of such information as the web can provide. Creating a meaningful view of the information is another beast entirely. Consider the Model-View-Control paradigm, pervasive in the domain of computer applications. This pattern mandates a strict decoupling of a Model and a View. Traditionally the Model is defined as that part of a system or application in which resides the "Business Logic" of the system or application. More often than not the data which the application uses, if any, is presented as part of the Model, though additional design patterns may endeavor to abstract it otherwise. The View then is that thing which is used to display that which is in the Model in such a way as is appropriate for the intention of the application. Applying this notion to the Semantic Web results in the resources of the Semantic Web being cast to the domain of the Model. The View then, highly variable based on the intended application, is that which presents resources to the intended consumer.
Some brief meditation is needed in order to see the consequence of this and the burden undoubtedly put upon the unfortunate fellow of whom a View is asked. It is easy enough to conceive of a relatively straight forward system designed for the purpose of viewing a single resource at a time. It is almost equally easy to do the same for a system providing a mapped view of the resources and their connections. Additional, and dare I say more realistic, requirements may place further restrictions on the View however. While I am not in a position to postulate a fully enumerated list of such requirements one can surmise that it may be of value to only show certain types of relationships. One may also surmise that, as we are considering a View, the consumer of said View may require a specific layout to be adhered to or multiple layouts depending on the type of resource presented. Such an implementation is cumbersome even once and the dynamic nature of the Semantic Web increases the burden on the developer of the View as they will most certainly be required to make constant updates to the view in keeping with the ever changing, ever growing, ontology space of the Semantic Web.
And thus we come to the very heart of the matter. Problem: A View of the Semantic Web must be capable of keeping pace with the Semantic web itself. Such a View must also allow for conformity to requirements concerning resource View positioning, layout, and associated resource requests, without recourse to re-implementation upon a change in the ontology.
~~
My efforts to be succinct in exposition have clearly failed. However, with problem stated progress can be made towards a solution with clarity.
DnL8Tar
-PCM
No comments:
Post a Comment