Domain Specific Program Synthesis

Program Synthesis refers to the task of constructing a program in a specific programming language, given its intent in a particular format. This emerging field can be applied in diverse domains and is currently being investigated with different techniques. A program synthesizer would simplify the efforts of programmers and help them focus on the program's core logic, without worrying about language syntax and other domain specifics. We applied the concepts of program synthesis in the context of solving a propositional logic word problem. We have developed a tool that is capable of understanding, parsing and evaluating a propositional logic word problem. With the user's natural language input, this tool processes the query and evaluates truth values of the question expressions. The working of the tool can be explained in three major phases: natural language processing, machine learning to obtain postfix notations of the Boolean expressions involved, and further evaluation of the postfix notations to determine the answers. Our goal was to explore the domain agnostic capabilities of our program-synthesis-based techniques of learning used in the implementation of this tool.