User Feedback-Driven Generative Models-Based Methodology for Service Construction

Software services are dynamic and can be composed or mashed up to create applications, with user feedback driving continuous optimization. In data-driven intelligent software development, supporting the composition and optimization of these services based on user–developer interaction data is a key challenge. Issues such as inaccurate service recommendations, unstructured feedback, and the ongoing evolution of services complicate this process. This article proposes a solution that leverages software service recommendations to construct mashups, extracts valuable insights from user feedback, and optimizes services in response. Optimization is addressed at both the service and code levels to alleviate developer workload. A case study demonstrates the service construction process, illustrating how these methods improve service composition and performance.