[27]                               home                          [29]

4/22/2004 6:53 AM

 

Key questions on Common Upper Ontology

 

 

Dean,

 

Before saying anything else, I want to emphasize that I am in favor of R & D work on developing ontologies.  But I also want to warn people not to expect them to be a panacea that will magically solve all (or even many) of their current problems.

 

My greatest fear about the semantic web is that the bubble will burst, and all the overhyped fever about it will create a backlash against any R & D that shares any common buzzwords with it.  That kind of backlash has occurred many times during the past 40 years, and many very good projects were canceled because they were tainted by the hype associated with some spectacular failures that richly deserved to fail.

 

Your recent note indicates some misconceptions that are widely shared, and which will inevitably lead to disillusionment:

 

However, the web (and in particular, web information systems like wikipedias and blogs) have shown us that there is a  large amount of information that is being built up in advance  of, or in the absence of, such an agreement, and while we  would love to be able to impose some controlled vocabulary  on these contexts, it is too late for some and socially  inappropriate for others.

 

The greatest misconception is the notion that "controlling vocabulary" is (1) possible, (2) desirable, and (3) capable of making some still undefined and unimplemented programs interoperable.  That hope is doomed to failure.

 

What can be done is to continue developing standards of the sort that have been successful in the past:  standards for floating-point numbers, programming languages, APIs, times, dates, measures, weights, currencies, etc.  These have been useful, and their continued development and standardization will be helpful for many reasons.

 

But the hope that the vocabulary used for blogs, wikipedias, or web sites in general can or should be controlled is not only doomed to failure, it is totally and absolutely wrong headed.

 

I have no fear that such control will ever be imposed, because it is impossible to do so.  However, I do fear that some people think that such control is desirable and that they might see the failure to establish it as a reason for killing R & D budgets for some very important work.

 

WordNet and other lexical resources that are available on the web, for example, are extremely valuable.  But they do not "control" how language should be used, but "describe" how people actually use language.

 

Summary:  Description is good, desirable, and possible. Control is impossible, and any attempt to enforce it is doomed to failure.  But the inevitable failure of attempts to control vocabulary may cause other highly desirable descriptive work to be canceled.

 

John Sowa