Revisiting Principles and Challenges in Natural Language Programming

Automation has faced the risk of reducing its pace due to the shortage of information technology professionals. Although part of the programming demand can be addressed by simple compositions of high-level functions and data flows, non-technical professionals are still unable to build their own software given the intrinsic complexity of coding. Among other types of end-user development, natural language programming has emerged as a strong candidate to fill this gap by developing methods and tools to allow end users to program. The paper revisits some principles of evaluation of traditional programming languages and analyses the new challenges to deliver an effective end-user development platform
based on aspects of natural language processing, human-computer interaction, software engineering, and programming education. We advocate that an effective end-user platform is essentially hybrid, combining features from different branches of the end-user development research, having, however, a search mechanism with semantic capabilities at its centre.