Applying ChatGPT-Based Iterative Improvement Model for Improving Software Maintenance Efficiency

Software maintenance takes a lot of manpower and cost. However, it is impossible to effectively improve the quality and efficiency of software maintenance. The biggest problem that maintenance personnel encounter is the lack of correctness, completeness, and consistency of documents. In addition, legacy systems generally lack refactor features of design architecture. These drawbacks are critical factors to cause more time and cost of software maintenance. ChatGPT has a strong analysis capability to provide positive and effective suggestions. Iterative training maintenance concept and technology to chatGPT then requires chatGPT to identify defects of incorrect, inconsistent, and incomplete contents of documents and suggests revision manners of design. With chatGPT, maintenance personnel modify the defects of artifacts and adjust the design architecture quickly and completely. However, it is impossible to completely improve software maintainability in once or twice modifications. Based on chatGPT we propose an Iterative Improvement Model (IIM) with mutation testing to evaluate the improvement effect of software maintainability in this paper. Iteratively feedback and train chatGPT to gradually eliminate the maintenance quality defects and problems. Based on chatGPT, an iterative improvement model can increase software maintainability gradually and improve software maintenance efficiency.