Subsumption provides a syntactic notion of generality. Generality can simply be defined in terms of the cover of a concept. That is, a concept, C, is more general than a concept, C′, if C covers at least as many examples as C′ (see Learning as Search). However, this does not tell us how to determine, from their syntax, if one sentence in a concept description language is more general than another. When we define a subsumption relation for a language, we provide a syntactic analogue of generality (Lavrač & Dčeroski, 1994). For example, θ-subsumption (Plotkin, 1970) is the basis for constructing generalization lattices in inductive logic programming (Shapiro, 1981). See Generality of Logic for a definition of θ-subsumption. An example of defining a subsumption relation for a domain specific language is in the LEX program (Mitchell, Utgoff, & Banerji, 1983), where an ordering on mathematical expressions is given.
- Mitchell, T. M., Utgoff, P. E., & Banerji, R. B. (1983). Learning by experimentation: Acquiring and refining problem-solving heuristics. In R. S. Michalski, J. G. Carbonell, & T. M. Mitchell (Eds.), Machine learning: An artificial intelligence approach. Palo Alto: Tioga.Google Scholar
- Plotkin, G. D. (1970). A note on inductive generalization. In B. Meltzer & D. Michie (Eds.), Machine intelligence (Vol. 5, pp. 153–163). Edinburgh University Press.Google Scholar
- Shapiro, E. Y. (1981). An algorithm that infers theories from facts. In Proceedings of the seventh international joint conference on artificial intelligence, Vancouver (pp. 446–451). Los Altos: Morgan Kaufmann.Google Scholar