Journal of Software-Evolution and Process最新文献

The cross-project defect prediction (CPDP) in software applications is crucial to predict defects and ensure software quality. The performance of the traditional CPDP models is degraded due to the class imbalance issue between different projects and differences in the data distribution. To overcome these limitations, a novel approach is proposed named as Levy flight–enabled greylag goose optimized UniXcoder-based stacked defect predictor (LFGGO-USDP) for the prediction of cross-project defects in the software engineering. In this paper, 23 software projects are selected from diverse datasets such as PROMISE, ReLink, AEEEM, and NASA that are preprocessed for enhancing reliability and reducing class imbalance issues. The transformation model maps source and target projects that are present in the feature space for enhancing predictive performances. During feature selection, the LF mechanism is embedded with the GGO algorithm to localize the features in the source code for enhancing diversity and minimizing local optimum issues. The integration of UniXcoder-based stacked bidirectional long short-term memory (U-SBiLSTM) is implemented as a cross-project defect predictor. The UniXcoder model extracts semantic information for source code tokenization. Then, the output of UniXcoder is fed as input to SBiLSTM, and the SBiLSTM model is applied to determine the relationship between the source code. After that, the output of UniXcoder (which contains the semantic features) is integrated with the output of SBiLSTM (which contains the sequential and temporal dependencies). After concatenating these features, the particular information is selected by using an attention mechanism for categorizing defective and nondefective classes. The experimental investigations are performed to analyze the nondefective and defective cases in software projects and numerical validation is conducted by applying different evaluation models for analyzing the superiority. The proposed model achieved the highest defect prediction accuracy of 0.986 compared to other existing approaches that demonstrates the proposed model provided better prediction outcomes.

The use of embedded systems has increased significantly over the last decade with the proliferation of Internet of Things technology, automotive and healthcare innovations and the use of smart home appliances and consumer electronics. With this increase, the need for higher quality embedded systems has increased. There are various guidelines and standards, such as ISO/IEC 9126 and ISO/IEC 25010, for product quality evaluation. However, these guidelines cannot be directly applied to embedded systems due to the nature of these systems. Applying traditional quality standards or guidelines on these systems without modification may degrade the performance of the system, increase memory usage or energy consumption, or affect other critical physical metrics adversely. Consequently, several models and approaches have either been introduced or have adopted existing guidelines to produce high-quality embedded systems. With this motivation, to understand the state of the art, and to identify the research directions in the field, we conducted a systematic literature review (SLR). In our research, we have investigated studies published from 1980 to 2024 and provided a comprehensive review of the scientific literature on quality models, quality attributes, employed practices, and the challenges, gaps, and pitfalls in the field.

Effective collaboration among software developers relies heavily on their ability to communicate efficiently, with emotions playing a pivotal role in this process. Emotions are widely used in human decision-making, making automated tools for emotion classification within developer communication channels essential. These tools can enhance productivity and collaboration by increasing awareness of fellow developers' emotions. Previous approaches, such as HOMER, RAKEL, and EmoTxt, have been proposed to classify emotions in Stack Overflow and Jira datasets at a finer granularity. However, these tools face performance challenges. To address these limitations, we aim to enhance multilabeled emotion classification performance by leveraging TextCNN, word embeddings, and hyper-parameter optimization. We validate the performance of this method by comparing it with the best previous methods for emotion classification in software engineering text. This approach achieves an F1-Micro score of 84.6001% on the Jira dataset and 76.9366% on the Stack Overflow dataset, showing an improvement of 3.5001% and 8.6366%, respectively. This advancement underscores the potential of this method in improving emotion classification performance, thereby fostering better collaboration and productivity among software developers.

The relationship between test code and production code, that is, test-to-code traceability, plays an essential role in the verification, reliability, and certification of software systems. Prior work on test-to-code traceability focuses mainly on Java. However, as Python allows more flexible testing styles, it is still unknown whether existing traceability approaches work well on Python projects. In order to address this gap in knowledge, this paper evaluates whether existing traceability approaches can accurately identify test-to-code links in Python projects. We collected seven popular Python projects and carried out an exploratory study at both the method and module levels (involving a total of 3198 test cases). On these projects, we evaluated 15 individual traceability techniques along with cross-level information propagation and four combining resolution strategies. The results reveal that the performance of test-to-code traceability approaches on Python has many differences with Java: (1) most of the existing techniques have poor effectiveness for Python; (2) after augmenting with cross-level information, the recall surprisingly drops; and (3) machine learning based combination approach achieves the best recall but the worst precision. These findings shed light on the best traceability approaches for Python projects, and also provide guidelines for researchers and the Python community.

Transformer methods have revolutionized software requirements classification by combining advanced natural language processing to accurately understand and categorize requirements. While traditional methods like Doc2Vec and TF-IDF are useful, they often fail to capture the deep contextual relationships and subtle meanings inherent in textual data. Transformer models possess unique strengths and weaknesses, impacting their ability to capture various aspects of the data. Consequently, relying on a single model can lead to suboptimal feature representations, limiting the overall performance of the classification task. To address this challenge, our study introduces an innovative BET-BiLSTM (balanced ensemble transformers using Bi-LSTM) model. This model combines the strengths of five transformer–based models BERT, RoBERTa, XLNet, GPT-2, and T5 through weighted averaging ensemble, resulting in a sophisticated and resilient feature set. By employing data balancing techniques, we ensure a well-distributed representation of features, addressing the issue of class imbalance. The BET-BiLSTM model plays a crucial role in the classification process, achieving an impressive accuracy of 96%. Moreover, the practical applicability of this model is validated through its successful implementation on three publicly available unlabeled datasets and one additional labeled dataset. The model significantly improved the completeness and reliability of these datasets by accurately predicting labels for previously unclassified requirements. This makes our approach a powerful tool for large-scale requirements analysis and classification tasks, outperforming traditional single-model methods and showcasing its real-world effectiveness.

The GitHub issue community serves as the primary means for project developers to obtain information about program bugs, and numerous GitHub users post issues based on encountered project vulnerabilities or error messages. However, these issues often vary in quality, leading to a significant time burden on project developers. By collecting 2500 bug-related issues from five GitHub projects, we first manually analyze a large volume of issue information to formulate rules for identifying whether a bug-tagged issue is truly fixed by project developers. We find that a substantial number (ranging from 29% to 68.4% in different projects) of bug-tagged issues are not truly fixed by project developers. We empirically investigate the characteristics of such issues and summarize the reasons why they are not fixed. Then, we propose an automated approach called DFBERT to identify the bug-tagged issues that are more likely to be fixed by project developers. Our approach incorporates both text and non-text features to train a neural network-based prediction model. The experimental results show that our approach achieves an average F1-score of 0.66 in inter-project setting, and the F1-score increase to 0.77 when adding part of testing data for training.

GUI ripping explores the graphical user interface of an application to build a model which can express the application behavior. The ripped GUI model is useful in various software engineering tasks. Traditional GUI ripping techniques depend on the underlying GUI frameworks to provide the GUI structure information. They are difficult to work across platforms or on nonnative applications where the GUI structure information cannot easily be obtained. This work introduces cvrip, a visual GUI ripping framework, to address the problem. cvrip visually analyzes the GUI screen for ripping and does not rely on the underlying GUI frameworks. We introduce many new techniques to enable efficient visual GUI ripping, for example, a YOLO v5-based model to detect executable widgets, a state recognition acceleration method for fast model updating, and several GUI exploration strategies taking the characteristics of imperfect visual analysis into account. Experiments are conducted to evaluate many technique choices in visual GUI ripping and compare the solution with the traditional style ripping. The results show that cvrip can get competitive exploration coverage compared to traditional approaches. This suggests visual GUI ripping is a direction worthy of more future studies.

Organizations use key performance indicators (KPIs) to assess the effectiveness and efficiency of their procedures and processes. In a world that is constantly evolving and hyperconnected via the internet, it is of great interest to analyze how changes (organizational, legal, technological or other) can lead to modifications in the KPIs involved. However, little attention has been paid to KPI evolution either in the scientific literature or in developed solutions. This paper presents A Flexible Framework for the Evolution, Consistency and Traceability of KPIs (AFFECTK) that aims at establishing the basis for suitable KPIs' evolution management. The feasibility of this proposal is demonstrated through a proof-of-concept developed using a reasoning tool based on Constraint Logic Programming. The framework is further evaluated, using real KPI case studies, to assess the functional suitability of our approach.