Randomizing the knowledge acquisition bottleneck

This paper addresses the knowledge acquisition bottleneck. It first takes an information-theoretic view of knowledge acquisition as having a basis in randomization theory and subsequently gives practical examples. The examples are taken from the field of expert compiler technology. Such technology can be used to effect the realization of fourth generation languages. These languages have been shown to be among software engineering´s greatest success stories. Expert compilers have the advantage of being easily maintainable and extensible. They randomize translational information in the form of rules. The capture of domain-specific knowledge allows for the construction of context-sensitive languages that are easy to work with. Of course, such languages are necessarily domain-specific; but even here expert compilers lend their advantage of promoting rapid prototyping within similar domains. It follows that if one builds a text-oriented domain-specific language that one can build a more complex expert compiler and so on. Clearly, randomization has application to not just the data, but to the representation of the data as well