A Supervised Generative Encoder-Based Approach to Improve Test Case Prioritization in Web Applications

An advanced software systems have to accomplish a
huge set of necessities because of their durability and complexity.
Accordingly, the essential stage of software testing in a software
engineering project with respective program behavior to its
requirements is evaluated. Regression testing is the process of
retesting the software system after variations have arisen, for
example, after a new form is emerged. Generally, only the subset
of test conditions are performed for a specific version as a result
of limited resources. To decrease these problems, one of the
method is to assume test case prioritization (TCP) while many
works has specified that TCP enhance an overall performance of
software testing. TCP hold numerous types of models, which
contains own weaknesses and strengths. Many studies have shown
an attention in the application of machine learning in TCP
recently. This study presents a Leveraging Artificial Intelligence
for Enhancing the Effectiveness of Test Case Prioritization
Techniques (LAI-ETCPT) in Web Application Testing. The main
objective of the LAI-ETCPT technique is to significantly enhance
software testing efficiency by effectively prioritizing critical test
cases. In the data preprocessing stage, the LAI-ETCPT method
handles missing values and uses a normalization technique to
ensure high-quality input data for further processing. For
dimensionality reduction, a hybrid approach combining minimum
redundancy maximum relevance and recursive feature
elimination is employed to recognize the most informative
features. In the classification process, a supervised variational
autoencoder is applied to capture complex relationships and
generate meaningful representations of test cases. To ensure an
improved performance of LAI-ETCPT model, a comprehensive
simulation analysis is conducted, and the results demonstrate the
improvement of the LAI-ETCPT technique over recent methods.
Keywords— Test Case Prioritization; Software Testing;
Minimum Redundancy Maximum Relevance; Supervised
Variational Autoencoder; Machine Learning