Automatic Generation of Algorithm Procedures using General Pre-trained Models

To create programs, it is essential to have a logical thinking skill that breaks down program purposes into the necessary processes. Until now, we had developed procedure learning systems that cultivate logical thinking namely ability to think with algorithms. In problem creation for procedure learning systems, we need to create a large quantity of procedures that are broken down into necessary smaller problem statements. In this research, we propose a method of generating program procedures from problem statements using a general pre-trained learning model.