IET Software最新文献

Currently, with the development of the financial market and the increase of personal credit products, the phenomenon of personal loan default has also attracted widespread attention. How to make effective predictions of loan default has become a research hotspot and difficulty in this field. This article addresses the problems in existing loan default prediction methods. Based on the domain knowledge of consumer loan defaults and machine learning theory, it is proposed to use a hybrid sampling method based on Mahalanobis distance SMOTEENN to balance the dataset. The sampling method is further extended from the original credit card fraud research to the study of loan default prediction. The balanced dataset is used to integrate categorical boosting (CatBoost)-GRU model with Bagging for consumer loan default prediction. First, the loan dataset of Lending Club platform is selected for default prediction experiments. Then, to further validate the effectiveness of the method, the method is applied to the loan default dataset of Kiva for further validation. Finally, through the comparison of the experimental results, it is proved that the method based on Mahalanobis Distance SMOTEENN hybrid sampling and Bagging integrating CatBoost-GRU can be very effective in loan default prediction. The method has strong application potential and practical effects in loan default prediction research, which not only improves the accuracy and efficiency of default prediction but also provides a wide range of insights and methodological references for the solution of similar problems.

Quantum computing is one of the research areas progressing rapidly toward practical deployment, yet the engineering of scalable and reliable quantum software remains underdeveloped. Current quantum software engineering (QSE) practices are largely tools-driven and ad hoc that providing limited support for managing probabilistic execution, hybrid quantum–classical workflows, noise sensitivity, and hardware constraints. This study proposed a structured QSE lifecycle that integrates quantum-specific characteristics with disciplined software engineering practices and principles. The proposed lifecycle organizes development into six phases, encompassing quantum requirements engineering, formal modeling, architecture and circuit design, hybrid integration, noise-aware testing, and deployment with monitoring. Each phase is supported by explicit artifacts and quantitative criteria to enable systematic progression and iterative refinement. The QSE is validated through expert assessment and simulation-based experimentation using representative variational quantum algorithms under the realistic noise conditions. The results show improved fidelity convergence, reduced resource overhead, enhanced development stability (DS), and more reliable validation compared with unstructured workflows, demonstrating the value of lifecycle-driven engineering for quantum software systems.

This study uses advanced approaches on the enlarged BRATS dataset to increase brain magnetic resonance imaging (MRI) image reconstruction accuracy and reliability. This study addresses MRI image processing issues such as noise, artifacts, and high-quality reconstruction. These traits are essential for brain tumor detection and analysis. This effort aims to establish a comprehensive image processing pipeline that standardizes MRI images, reduces noise, and improves clarity for better image reconstruction. ESORecon-Net, which combines the echo state network (ESN) and osprey optimization algorithm (OSPREY), manages raw k-space data cleverly and improves reconstruction. The model’s dual-phase optimization ensures accuracy and efficiency in reconstructing high-quality MRI images. The proposed ESORecon-Net achieved a peak signal-to-noise ratio (PSNR) of 49.12 dB and a structural similarity index measure (SSIM) of 0.993, surpassing existing methods such as the fully sampled k-space-trained network (FS-kNet) and motion-informed deep learning network (MIDNet). These results confirm ESORecon-Net’s effectiveness in enhancing brain MRI image reconstruction, improving both image quality and computational performance.

Typical source code (SC) metrics are useful in identifying and predicting the used design patterns in typical Java and Kotlin projects. However, typical SC metrics–based prediction tends to be less accurate. This research presents a novel idea of detecting various design patterns in a code with the help of large language model (LLM)–based features extraction, instead of using conventional SC metrics typically used in the existing approaches. This research aims to identify and extract using architectural design patterns with the help of various LLM–based feature extraction and supervised machine learning (ML) algorithms. In the proposed approach, LLM–based various 24 design pattern features are extracted instead of the start-of-the-art metrics used for prediction of the design pattern of a particular SC of project. This paper mainly contributes to intelligent and automated software design and development in terms of artificial intelligence (AI)–based detection of design patterns for the purpose of reengineering. In addition to this, this research also aims to investigate the role of design patterns features in automated detection of architectural design patterns and study the association in architectural design patterns and its respective and peculiar features. A Python-based implementation of support vector machine (SVM) algorithm was made. The overall accuracy of SVM–based binary classification was 97.30% that guides the performance of the proposed approach.

Traditional higher education faces significant challenges, including rising costs, inflexibility, and a disconnect from workforce demands, while current credentialing systems are often centralized and vulnerable to fraud. Microcredentials have emerged as a solution, yet they lack integration into formal degree pathways and face issues with recognition and security. This paper introduces a blockchain based framework designed to securely issue, manage, and verify credit-bearing microcredentials, bridging the gap between non-traditional learning and formal academic programs. The proposed system utilizes a hybrid on-chain/off-chain architecture, leveraging Hyperledger Besu, smart contracts, and the InterPlanetary File System (IPFS) for decentralized storage. The system’s smart contracts automate the entire credential lifecycle, including issuance, revocation, verification, and retrieval. Moreover, this research presents a comprehensive performance evaluation using custom scripts and Hyperledger Caliper which confirmed the system’s operational feasibility. The system achieved a stable and predictable throughput of 1.6–2.0 transactions per second (TPS) under heavy loads, with an average latency between 0.93 and 4.34 s and a mean of 1.88 s. These findings affirm that the architecture provides a robust and responsive solution for academic credentialing. Overall, this study presents a practical and scalable framework that enhances the trust, portability, and integration of microcredentials within higher education. By enabling secure, learner-owned, and verifiable records, the system offers a trusted pathway for recognizing prior learning and streamlining academic progression.

The modern metropolis essentially demands the use of state-of-the-art, real-time surveillance systems, which should be reliable, scalable, and respectful of privacy at the same time. Critical shortcomings in traditional architectures are single points of failure, poor scalability, frequent data breaches, and inadequately managed privacy. These aspects of themselves make it inept for the demands of dynamic, fast-paced city environments, without which reliability, security, and adaptability cannot be compromised at any cost. This brings to light the critical need for innovative and decentralized solutions that can overcome these challenges comprehensively. In our proposed approach, a decentralized framework integrates private blockchain technology via Ethereum, a hybrid cryptography model combining advanced encryption standard (AES) and Rivest–Shamir–Adleman (RSA) encryption, and state-of-the-art deep learning techniques such as YOLOv8, DeepSort, and ArcFace. Blockchain technology ensures metadata is immutable and transparent, thus saving metadata from unauthorized access and tampering. The hybrid cryptography model encrypts sensitive data through AES and securely shares the key of AES through RSA encryption, while decryption is efficiently done in a key management system (KMS). Furthermore, YOLOv8 and DeepSort can be used for high-precision object detection and real-time tracking, and ArcFace can be used for facial recognition, meeting the split-second decision-making required in urban surveillance. Extensive experiments are performed, and the results indicate that the proposed framework enhances detection precision, tracking accuracy, real-time responsiveness (60 FPS), and resistance to tampering (>99% chain quality per quorum Byzantine fault tolerance [QBFT]) without compromising efficiency. The adaptive and reliable solution meets modern urban surveillance demands that are evolving at an ever-increasing pace. The scalability of the operation further ensures enhanced public safety. This paper discusses a decentralized urban surveillance system that is both tamper-proof and secure using current blockchain technologies, InterPlanetary file system (IPFS), hybrid AES–RSA, and deep learning technologies to mitigate the risks of a traditional centralized system, such as data tampering and privacy violations. The system uses the Ethereum blockchain to provide immutable metadata, the IPFS protocol to create a fully distributed storage system of video and image frames, and an off-chain KMS service to distribute the keys to the authorized edge devices. The system utilizes real-time object detection (YOLOv8), tracking (DeepSort), and face recognition (ArcFace) to perform inference locally on the edge devices. We have performed experiments that demonstrate the tamper-proof and secure scalability with low latency and secure tamper-proof data integrity of this urban surveillance system in ever-changing urban environments.

Open-source libraries are indispensable for modern software development but can create substantial maintenance burdens when they become deprecated or unmaintained. Selecting an appropriate replacement among many candidates remains challenging, since methods relying only on historical mining or similarity metrics often miss subtle differences in meaning. We propose AMRerank, a novel framework that integrates multi-agent qualitative analysis with a data-driven, interpretable reranking model. AMRerank first deploys specialized agents to examine and classify semantic relationships between libraries, generating evidence-backed labels and concise summaries. An interpretable reranking framework then fuses these qualitative signals with heuristic and semantic features to produce a fine-grained, explainable ranking. Evaluated on the GT2014 benchmark against competitive baselines (LMG, MMR, MMRLC), AMRerank achieves Precision@1 of 0.899 and mean reciprocal rank (MRR) of 0.928. As our case studies show, the system provides actionable, human-readable evidence that helps developers make more reliable migration choices.

The ability to comprehend complex viewpoints in text is critical for sentiment analysis (SA), particularly at the aspect level, yet existing models struggle with accurately identifying sentiment polarities and aspect-specific expressions due to their reliance on large, manually annotated, domain-specific datasets. To address these challenges, this paper introduces hybrid deep learning and RoBERTa-based SA (HDR-SA), a novel hybrid deep learning framework that integrates convolutional neural networks (CNNs), bidirectional long short-term memory (BiLSTM) networks, and the RoBERTa transformer model to perform comprehensive sentiment and aspect analysis. The proposed model begins with rigorous data preprocessing and normalization, utilizes Valence Aware Dictionary and sEntiment Reasoner (VADER) for sentiment scoring, constructs embedding vectors via Word2Vec, and employs a CNN-BiLSTM architecture enhanced by RoBERTa to capture both sequential and contextual embeddings for refined sentiment classification. The novelty of HDR-SA lies in its hybrid integration of conventional natural language processing (NLP) techniques with deep learning and transformer-based contextual understanding, enabling robust SA without the extensive need for domain-specific annotated data. Evaluated on the large-scale 515K Hotel Reviews dataset, HDR-SA achieved an accuracy of 95.75%, a precision of 0.96, a recall of 0.97, and an F1-score of 0.96, outperforming contemporary models such as target-dependent LSTM (TD-LSTM), ResNet-SCSO, and CNN-GA. These results demonstrate HDR-SA’s effectiveness in aspect-level SA and its scalability across diverse domains while reducing dependency on annotated resources.

Directed acyclic graph (DAG)-based blockchain is a promising paradigm of a blockchain system. During DAG-based consensus processes, nodes generate blocks in parallel, as well as voting YES on the previous blocks. It makes the DAG-based blockchain a performance advantage in terms of confirmation delay and transaction throughput. However, Byzantine faults affect DAG-based blockchain performance by not voting on blocks. The related analysis has not been explored. To this end, based on the most typical DAG-based consensus mechanism with Byzantine Fault Tolerance, hashgraph, we investigate the resilience of the DAG-based blockchain when Byzantine faults do not vote on blocks. First, we propose differential equations to model the running processes of the DAG-based blockchain in both high-load and low-load networks. It reveals the impact of Byzantine nonvoting behaviors on blockchain performance in a mathematical manner. Second, Byzantine nonvoting adversaries can leave the target node orphaned in an opportunistic network of low-load regimes. We propose a ranger-assisted DAG-based blockchain to alleviate the problem. It employs a group of rangers to collectively commit blocks and introduces the reputation, selection probabilities, and shuffling of rangers to supervise node behaviors. The performance of the proposed blockchain is also quantitatively analyzed. Third, we develop a blockchain simulator. The numerical results indicate the validity of the proposed analysis and the efficiency of the proposed blockchain.

Tourism is a global industry that increasingly relies on web-based applications to provide tourists with information about destinations, routes, food, accommodation, and transport facilities. To enhance the personalized experience for tourists using these platforms, incorporating a recommendation component is essential. However, tourism recommendations face significant challenges, particularly data sparsity and the cold start problem, which can negatively impact the accuracy of suggestions. This study introduces an innovative hybrid recommender system designed to deliver personalized travel experiences. To address the issues of data sparsity and cold start, the proposed recommender system employed a combination of several filtering techniques. The fusion of these techniques improves user satisfaction by providing accurate and diverse travel recommendations. The proposed system has been specifically designed for the local tourism landscape. The dataset utilized in this study was collected from TripAdvisor to evaluate the system’s performance. The results indicate that the hybrid recommender system achieves high accuracy, with an accuracy rate of 90.71%. Compared to previous studies, the proposed approach significantly improves the delivery of personalized travel recommendations. The findings highlight the effectiveness of combining multiple filtering techniques to generate precise and diverse suggestions tailored to user preferences.