Code Generation by Example Using Symbolic Machine Learning

Abstract Code generation is a key technique for model-driven engineering (MDE) approaches of software construction. enables the synthesis applications in executable programming languages from high-level specifications UML or domain-specific language. Specialised code and tools have been defined; however, task manually constructing generator remains substantial undertaking, requiring high degree expertise both source target languages, In this paper, we apply novel symbolic machine learning techniques tree-to-tree mappings syntax trees, to automate development generators source–target example pairs. We evaluate approach on several tasks, compare other construction approaches. The results show that can effectively examples, with relatively small manual effort required compared existing also identified it be adapted learn abstraction translation algorithms. paper demonstrates applied assist manipulating trees.