2021/05/20 by Ritu Kapur, Sumit Kalra, Kapur, Ritu +5
Computer Science · #Advanced Software Engineering Methodologies #FOS: Computer and information sciences #Software Engineering (cs.SE) #Software Engineering Research #Software Engineering Techniques and Practices #Software System Performance and Reliability
paper · pdf · doi:10.48550/arxiv.2105.09595
openalex publication_date 2021/05/20 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
A software architect uses quality requirements to design the architecture of\na system. However, it is essential to ensure that the system's final\narchitectural design achieves the standard quality requirements. The existing\narchitectural evaluation frameworks require basic skills and experience for\npractical usage, which novice software architects lack.\n We propose a framework that enables novice software architects to infer the\nsystem's quality requirements and tactics using the software architectural\nblock-line diagram. The framework takes an image as input, extracts various\ncomponents and connections, and maps them to viable architectural patterns,\nfollowed by identifying the system's corresponding quality attributes (QAs) and\ntactics. The framework includes a specifically trained machine learning model\nbased on image processing and semantic similarity methods to assist software\narchitects in evaluating a given design by a) evaluating an input architectural\ndesign based on the architectural patterns present in it, b) lists out the\nstrengths and weaknesses of the design in terms of QAs, c) recommends the\nnecessary architectural tactics that can be embedded in the design to achieve\nthe lacking QAs.\n To train our framework, we developed a dataset of 2,035 architectural images\nfrom fourteen architectural patterns such as Client-Server, Microservices, and\nModel View Controller, available at\nhttps://www.doi.org/10.6084/m9.figshare.14156408. The framework achieves a\nCorrect Recognition Rate of 98.71% in identifying the architectural patterns.\nWe evaluated the proposed framework's effectiveness and usefulness by using\ncontrolled and experimental groups, in which the experimental group performed\napproximately 150% better than the controlled group. The experiments were\nperformed as a part of the Masters of Computer Science course in an Engineering\nInstitution.\n