Providing a framework for determining important dependability parameters for web services using multi-criteria decision-making methods and neural networks

This paper highlights the importance of dependable web services, focusing on Google's offerings, and presents an advanced methodology for evaluating their significance through the selection of relevant dependability parameters. A robust framework combining multi-criteria decision-making (MCDM) techniques and neural networks is proposed, utilizing methods such as AHP, Shannon Entropy, RBF, SOM, MLP, TOPSIS, and COPRAS. The methodology is applied to assess six dependability indices—Availability, Safety, Reliability, Maintainability, Confidentiality, and Integrity—for three Google web services (Google Maps, Google Scholar, and Google Translate) in both general and healthcare contexts. Experimental results, including expert evaluations and data clustering, show that Shannon Entropy and RBF consistently outperform other methods in accuracy and adaptability. RBF excels in Availability (ranking 1.34) and Safety (1.67), while Shannon Entropy leads in Reliability (1.42), Maintainability (1.57), and Confidentiality/Integrity (1.63). Numerical analyses reveal key findings: LG category achieves the highest scores in Availability (1.0) and Integrity (1.0), while category TH (0.678) and TG (0.614) require significant improvements in Availability and Safety. In contrast, simpler methods like AHP and SOM were less effective, especially in handling complex data. This study emphasizes the importance of selecting evaluation methods suited to data characteristics, offering valuable insights for researchers and practitioners seeking robust evaluation techniques for web services, particularly in general and healthcare sectors.