Recommendation Framework for Blockchain-based Trustworthy Software Development

In the rapidly evolving business landscape, blockchain technology emerges as a key innovator, enhancing trust, transparency, and security. However, the unique features of blockchain pose challenges in developing blockchain-based software (BSD) systems, demanding improvements in conventional software development processes. Objective: This study aims to identify BSD process areas and develop a success probability prediction framework, enhancing BSD process success and progression. Method: We conducted a comprehensive literature survey and a questionnaire-based survey with practitioners to identify BSD process areas and gather training data. The study employs the Grey Wolf Optimizer (GWO) combined with the Naive Bayes Classifier to create a success probability prediction framework for BSD processes. Results: Our research identifies 47 BSD process areas, categorized across five software process improvement (SPI) stages: initial, managed, defined, quantitatively managed, and optimizing. The GWO algorithm facilitates the design of a predictive framework, assessing the success probability of each stage, encompassing various process areas. The framework also prioritizes process areas for each stage, helping practitioners identify critical areas considering implementation cost and success probability. Conclusion: Organizations using BSD can leverage this framework to improve their BSD processes. This study contributes to blockchain technology applications in software development, offering a systematic, predictive approach to augment the effectiveness and success rate of BSD processes.

在快速演进的商业格局中，区块链技术已然成为关键创新驱动力，显著提升了信任水平、透明度与安全性。然而区块链的独特特性为基于区块链的软件（Blockchain-Based Software，BSD）系统的开发带来了诸多挑战，亟需优化传统软件开发流程。

研究目标：本研究旨在明确BSD流程域，并构建一套成功概率预测框架，以提升BSD流程的实施成功率与推进效率。

研究方法：本研究通过全面的文献调研以及面向从业者的问卷调查，明确BSD流程域并收集训练数据集。本研究将灰狼优化算法（Grey Wolf Optimizer，GWO）与朴素贝叶斯分类器（Naive Bayes Classifier）相结合，用于搭建BSD流程的成功概率预测框架。

研究结果：本研究共识别出47个BSD流程域，可划分为软件过程改进（Software Process Improvement，SPI）的五个阶段：初始级、已管理级、已定义级、已量化管理级以及优化级。灰狼优化算法助力本预测框架的构建，可对覆盖各类流程域的各阶段成功概率进行评估；该框架还可对各阶段的流程域进行优先级排序，帮助从业者结合实施成本与成功概率，甄别关键流程域。

研究结论：采用BSD技术的组织可借助本框架优化其BSD开发流程。本研究为区块链技术在软件开发领域的应用提供了支撑，提出了一套系统化的预测方法，可有效提升BSD流程的实施效能与成功率。