Automated Control Logic Test Case Generation using Large Language Models

This work aims to generate White-box Test Cases from a mainframe NATURAL code fragment, using the Control Flow Graph Technique. Basically, it enables a code fragment analysis, generating its control flow represented by a Graph perspective. As a consequence, it provides the automatic generation of White-box Test Cases. Furthermore, this work contributes to attain a lower degree of difficult in the execution of White-box Tests within a Mainframe environment. Also, it brings a significant contribution related to the execution time, inherent to the software testing. At the end, it is added a testing expertise to NATURAL development teams. Usually there is not enough available time to test all possible paths in a algorithm, even in a simple application. Based on it, this work provides automatic generation of test cases for empowerment of team decision, regarding which test cases are more relevant to be performed.

Test cases are essential parts to perform software testing as they play very significant role defining the test steps to work. Test cases can be generated either manually or automatically. Creating test cases manually has several disadvantages, such as higher probability in producing inaccurate results. Furthermore, generating test cases manually is ineffective since manual methods will take longer and require high human efforts. Therefore, generating test cases automatically is more preferable to limit the previous problems. Automated testing can also provide more accurate results than manual testing because it is less prone to human errors encouraging software developers to generate test cases automatically. Test cases can be generated from various sources, including UML diagrams. In this paper, we propose a method to generate test cases automatically by utilising depth-first search algorithms. The UML diagram used in this study is an activity diagram. This paper also provides a method to parse activity diagram that is created using the visual paradigm application that has been converted into an XML format. Our method is able to parse and convert the XML format into graph or tree form. Using our proposed method, a test case search will be carried out using the depth-first search on the tree form to generate test cases. Lastly, we also evaluate our method by comparing the generated test cases with cyclomatic complexity, the method will be performed manually and automatically to validate the results of many scenarios. Our result showed that our method could generate the same scenario (100% coverage) as the manual validation method.

At present, the design of the test process during the manufacturing process of the aircraft is mainly performed by the process personnel to consult the relevant data of the tested object to obtain the test information, and manually write the test process file. Therefore, the writing efficiency is low and the writing quality relies heavily on the experience of craftsmen, and the standardization and standardization of process files cannot be effectively guaranteed. On the other hand, the test program is closely coupled with specific test equipment. The upgrade and update of test resources cause the test program to change, which increases the use and maintenance costs.This paper proposes to implement standardized modeling of test knowledge based on the Automatic Test Markup Language (IEEE 1671 ATML), to support the sharing, transplantation and reuse of the test information, and to standardize and digitize test process files. At the same time, the automatic generation of signal-oriented test cases is achieved, the interchangeability of test instruments and the portability of test programs are solved, and the universal execution of various test resources is achieved.

ABSTRACTCode generation for users' intent has become increasingly prevalent with the large language models (LLMs). To automatically evaluate the effectiveness of these models, multiple execution‐based benchmarks are proposed, including specially crafted tasks, accompanied by some test cases and a ground truth solution. LLMs are regarded as well‐performed in code generation tasks if they can pass the test cases corresponding to most tasks in these benchmarks. However, it is unknown whether the test cases have sufficient test adequacy and whether the test adequacy can affect the evaluation. In this paper, we conducted an empirical study to evaluate the test adequacy of the execution‐based benchmarks and to explore their effects during evaluation for LLMs. Based on the evaluation of the widely used benchmarks, HumanEval, MBPP, and two enhanced benchmarks HumanEval+ and MBPP+, we obtained the following results: (1) All the evaluated benchmarks have high statement coverage (above 99.16%), low branch coverage (74.39%) and low mutation score (87.69%). Especially for the tasks with higher cyclomatic complexities in the HumanEval and MBPP, the mutation score of test cases is lower. (2) No significant correlation exists between test adequacy (statement coverage, branch coverage and mutation score) of benchmarks and evaluating results on LLMs at the individual task level. (3) There is a significant positive correlation between mutation score‐based evaluation and another execution‐based evaluation metric () on LLMs at the individual task level. (4) The existing test case augmentation techniques have limited improvement in the coverage of test cases in the benchmark, while significantly improving the mutation score by approximately 34.60% and also can bring a more rigorous evaluation to LLMs on code generation. (5) The LLM‐based test case generation technique (EvalPlus) performs better than the traditional search‐based technique (Pynguin) in improving the benchmarks' test quality and evaluation ability of code generation.

A primary issue in black-box testing is how to generate adequate test cases from input domain of the system under test on the basis of user's requirement specification. However, for some types of systems including embedded systems, developing test cases from output domain is more suitable than developing from input domain, especially, when the output domain is smaller. This approach ensures better reliability of the system under test. In this paper, the authors present a new approach to automate the generation of test cases from output domain of a pilot project "Temperature Monitoring and Controlling of Nuclear Reactor System" (TMCNRS) which is an embedded system developed using modified Cleanroom Software Engineering methodology. An Automated Test Case Generator (ATCG) that uses Genetic algorithms (GAs) extensively and generates test cases from output domain is proposed. The ATCG generates test cases which are useful to conduct pseudo - exhaustive testing to detect single, double and several multimode faults in the system. The generator considers most of the combinations of outputs, and finds the corresponding inputs while optimizing the number of test cases generated. In order to investigate the effectiveness of this approach, test cases were generated by ATCG and the tests were conducted on the target embedded system at a minimum cost and time. Experimental results show that this approach is very promising.

Maximum coverage with minimum testing time is the main objective of a test case generation activity which leads to a multi-objective problem. Search-Based Testing (SBT) technique is a demanding research area for test case generation. Researchers have applied various metaheuristic (searching) algorithms to generate efficient and effective test cases in many research works. Out of these existing search-based algorithms, Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) are the most widely used algorithms in automatic test case generation. In this paper, a test case generation approach is proposed using Cuckoo Search (CS) algorithm. CS has a controlling feature, Levy flights, which makes it more efficient in searching the best candidate solution. It helps to generate efficient test cases in terms of code coverage and execution time. In our proposed method, test cases are generated based on path coverage criteria. Fitness of a test case is evaluated using branch distance and approximation level combined functions. The result is compared with PSO and with its variant Adaptive PSO (APSO). The experimental result shows that both the algorithms give nearly equal to the same result. Though the results are nearly equal, the implementation of CS is simple as it requires only one parameter to be tuned.

Testing of product is perform to discover or detect the errors and defects in the developed system. But testing is usually time consuming especially when complex projects are canvass. Testing of a product lead off with generation of test cases. The Test case generation are based on three parts coding, design and specification. The Specification based testing deals with generation of test cases from the functional requirements. The proposed system deals with automatic generation of test cases from functional requirement using Natural Language Processing (NLP). The proposed system constructs test cases based on keywords in context from functional requirement of Software Requirement Specification document. The proposed system is beneficial as it can automatically analyze the functional requirement from Software Requirement Specification in order to extract test cases for testing. The goal of the proposed system is to reduce effort and time consumed by software tester to test the product.

The vastly growing complexity of software enabled functionality – especially in the automotive industry where it is usually distributed over a network – requires also a steadily increasing number of test cases to ensure a high product quality. The automatic generation of test cases can ease this situation, but suffers from the well-known state space explosion problem, which is caused by the often infinite number of possible execution paths. The state space can be minimized by rules applied to test case generation leading to an enhanced scalability. A flexible rule adjustment according to the model structure allows a tailored test case generation capable of being fast as well as complete according to a given goal. In this work the implications of such rules on the searching capabilities are studied. The analysis is performed on a symbolic test generation engine based on a network of timed, deterministic UML state machines.

Services are highly reusable, flexible and loosely coupled, which makes the evolution and the maintenance of composite services more complex. Evolution of BPEL composite service covers changes of processes, bindings and interfaces. In this paper, an approach is proposed to select and generate test cases during the evolution of BPEL composite service. The approach identifies the changes by using control-flow analysis technique and comparing the paths in new composite service version and the old one using extensible BPEL flow graph (or XBFG). Message flow is appended to the control flow so that XBFG can describe the behavior of composite service integrally. The binding and predicate constraint information added in XBFG elements can be used in path selection and test case generation. Theory analysis and case study both show that the approach is effective, and test cases coverage rate is high for the changes of processes, bindings and interfaces.

Testing mobile application refers to the process of ensuring the quality and functionality of the software applications implemented for mobile phones. It is a fundamental step to test mobile applications specially before being released to the market. Load testing is a kind of software testing which is carried out to test the performance of mobile applications under different workloads. However, most of mobile apps are implemented within a bounded resource, budget, and time due to the market pressure. This extremely influences the quality of the mobile applications development industry. The automatic test case generation is the key factor for efficient and cost-effective software verification process. Model-based load testing represents both performance aspects and functional aspects of mobile applications by using modeling languages like unified modeling language (UML). Then, adopting the resulting models to automatically design test cases that will be used for load testing. This paper proposes an automatic model-based test case generation approach for mobile applications load testing. The proposed approach lessens testing time and satisfies fast time-to market constraint of mobile apps. Additionally, the proposed approach validates the mobile applications functionalities as well as the performance requirements.

Testing is the fundamental technique to assess the correctness of software systems, but it is cost-labored to generate test cases. One solution to change the situation is to automatize some parts of the testing process, especially the generation of test cases using formal theory and technology. The research work in the direction shows the good perspective. This paper targets on the automatic generation of test cases based on IOSTS, which is widely used to model reactive systems with data. When selecting test cases based on a set of test purposes specified by IOSTS or temporal logic, in general, the redundancy phenomena are unavoidable in the derived test suite. Hence, some strategies are presented for eliminating the redundancies in order to reduce the cost of implementing testing. More importantly, the strategies are directly applied to test cases in form of IOSTS, such that it can reduce not only the size of test suite, but also the cost of deriving test cases.

Extended finite state machine (EFSM) is among the most popular models for model-based testing. However, automated test case generation on EFSM models is still a challenging task since an EFSM model may contains infeasible paths. This paper proposed a novel approach (ATGEM) to generate test case and construct oracle information from EFSM automatically. To address the infeasible problem, a metric based on data flow analysis is presented to predict the infeasible probability so as to bypass the infeasible paths as far as possible and improve the test case generation efficiency. Afterwards, an executable EFSM model is developed to obtain runtime feedback information as a fitness function in order to generate test data and construct oracle information automatically. This approach, which can generate various types data and has a wide range of applications, combines static analysis and dynamic analysis aims to find a preferable feasible path subset to generate test cases and meet adequacy coverage criteria. The experimental results on several EFSM models show that test case generation method and path feasibility metric have good effectiveness. Utilizing path feasibility metric can speed up the process of test case generation greatly, and ATGEM is more efficient than existing method.

Background & Problem Statement - Software testing is a critical phase in the software development lifecycle (SDLC), ensuring that applications function correctly, meet user requirements, and maintain high- quality standards. Traditional software testing approaches, including manual testing and rule-based automation, often face challenges in scalability, efficiency, and adaptability to dynamic software environments. Traditional testing methods are overwhelmed by complex software systems which slows down defect detection and extends both testing costs and release schedules. Machine Learning (ML) has emerged as a transformative solution, introducing predictive and adaptive capabilities that optimize test case selection, automate defect detection, and enhance overall software quality assurance (QA). This study explores the integration of ML in software testing, addressing the challenges of traditional QA methodologies and demonstrating how AI-driven frameworks improve testing efficiency. Methodology - To investigate the impact of ML in software testing, this research adopts a systematic approach by analyzing ML-driven test automation techniques, including predictive testing, adaptive test execution, and automated test case generation. Research reviews how Google Microsoft Facebook IBM and Deep Code put ML-based quality assurance frameworks into operation. The study leverages supervised learning, reinforcement learning, deep learning, and NLP-based techniques to demonstrate how ML models predict software defects, dynamically adapt test cases, and optimize testing resources. The research tests how ML-based testing models operate within CI/CD pipelines to improve ongoing testing and deployment flow. Analysis & Results - The analysis of ML-driven software testing reveals that predictive analytics improves early defect detection rates. It helps developers spend 37% less time debugging their work. Adaptive testing models, including self-healing test scripts, minimize maintenance costs by 50% and enhance test reliability in agile environments. The integration of NLP-based test case generation increases test coverage. NLP technology enables automatic connection between requirements and test cases at 89% success rate. Additionally, reinforcement learning techniques improve test case selection, reducing redundant test executions by 43%. Our research shows different ML methods work well to lessen incorrect error alerts. ML integration for QA surely increasing defect prediction accuracy and optimizing test execution time. Findings & Contributions - This research contributes to the field of AI-driven software testing by providing a comprehensive framework for ML-based QA methodologies. Our study shows that machine learning helps find more software problems better adapts test cases and lowers testing expenses to solve present software development needs. The study also identifies critical challenges, including data availability, model interpretability, and computational overhead, suggesting future research directions in Explainable AI (XAI), hybrid AI-ML testing models, and AI-driven security testing. As the industry moves toward AI-first software testing, this research paves the way for fully autonomous QA frameworks, enabling intelligent, scalable, and cost- effective software validation techniques. Keywords - Machine Learning, Software Testing, Quality Assurance, Predictive Testing, Adaptive Testing, Test Automation, Defect Prediction, Self- Healing Test Scripts, AI-Driven QA, Reinforcement

Software systems are not considered complete unless properly tested and verified. In existing literature, a growing interest on establishment of automated testing techniques has been observed. However, tedious manual process of test case generation largely depends upon domain knowledge and formalized representation of user requirements. The advent of semantic web engineering has led the foundation for developing ontologies as a mean to express information and knowledge semantics regarding particular domain efficiently. In software testing, ontologies can be significantly helpful to automate testing phase as they encode domain knowledge in machine interpretable format. We have proposed automatic test case generation framework that involves ontology-based requirement specification and learning based methods for conducting black box testing. Our approach integrates knowledge-based system (ontology) with learning-based testing algorithm to automate: generation of test cases, test execution and test verdict construction. Proposed framework includes, requirement ontology to formalize requirement specification, Dialogue Manager that enables selection of available test cases and Learning Based Testing to generate counter examples of test cases through system learning. The contribution of this paper is to enable 1) requirement elicitation and specification using ontologies 2) test data selection from existing ontologies and 3) automatic test case generation from existing test cases. To represent the applicability of this research, ontology for requirement elicitation and specification is developed. Framework proposed in this research paper is an effort to provide software testing tools to save time, cost and efforts during test design phase.

Although Android apps are becoming more and more popular, apps tend to contain defects which can ultimately manifest as failures (or crashes) to end-users. In order to improve software quality, people have proposed different automated tools for testing Android applications. The automatic generation of test cases is an important means to improve the efficiency of software testing, but many tools related to test case generation do not seem to be fully adapted to the problem of Android test case generation. Therefore, this article aims to design and evaluate the automatic generation method of Android application test cases based on the multi-dimensional coverage. We conducted test experiments on 48 self-developed Android apps through the Gradle plugin, and the final average target coverage reached about 32%. Considering that the current component modules are not yet complete, we believe that tools will reach even higher coverage in the future.