Generational analysis of variety in data structures: impact on automatic data integration and on the semantic web

摘要

We examine data definition languages (DDLs) from various computing era spanning almost 50 years to date. We prove that contemporary DDLs are indistinguishable from older ones using Zipf distribution of words, Zipf distributions of meanings, and information theory. None addresses the Law of Requisite Variety, which is necessary for enabling automatic data integration from autonomous heterogeneous data sources and for the realization of the Semantic Web. The growth of the entire computing industry is hampered by the lack of progress in the development of DDLs suitable for these two goals. Our findings set the stage for the future development of a mathematically sound DDL better suited for the aforementioned purposes.