The problem of practically feasible inductive inference of functions or other objects that can be described by means of an attribute grammar is studied in this paper. In our approach based on attribute grammars various kinds of knowledge about the object to be foudn can be encoded, ranging from usual input/output examples to assumptions about unknown object's syntactic structure to some dynamic object's properties. We present theoretical resutls as well as describe the architecture of a practical inductive synthesis system based on theoretical findings.
展开▼
