ILP and Automatic Programming:
Towards Three Approaches
Pierre Flener, Bilkent University, Ankara, Turkey
Lubo\v{s} Popel\'{\i}nsk\'{y}, Masaryk University, Brno, Czech Republic
Olga \v{S}t\v{e}p\'{a}nkov\'{a}, CTU Prague, Czech Republic
The prospects of inductive logic programming (ILP) with respect to automatic
programming (program synthesis) are discussed. We argue that logic program
synthesis from incomplete information is but a niche of ILP, and study
consequences of this statement. Then, three
approaches are described: schema-driven synthesis of logic programs from
incomplete specifications, the role of transformation techniques in ILP, and
interactive assumption-based inductive learning.}