Expert Systems with Applications最新文献

{"title":"Cross-domain recommendation via interest-aware pseudo-overlapping user alignment","authors":"Shitong Xiao ,&nbsp;Rui Chen ,&nbsp;Hongtao Song ,&nbsp;Qilong Han","doi":"10.1016/j.eswa.2025.128638","DOIUrl":"10.1016/j.eswa.2025.128638","url":null,"abstract":"<div><div>The cold-start problem remains a classic challenge in recommender systems. Cross-Domain Recommendation, which utilizes information from auxiliary source domains to boost performance, presents an effective solution. Bridge-based cross-domain methods are especially beneficial for cold-start users, who have interactions in the source domain but not the target domain. These methods typically learn a mapping function to transfer user preferences from source to target domain. However, they face two significant challenges: (1) Dependency on overlapping users, as the mapping function’s training largely relies on the limited number of overlapping users available in practical scenarios. (2) The uniform user embeddings lack the capacity to reflect multiple interests of users in the target domain, leading to weak expression of mapped users. To tackle these challenges, we introduce a new cross-domain recommendation model. Initially, the model learns a global shared interest pool across domains using an interest activation network. It then groups users by their activated interests and matches them with pseudo-overlapping users within the same interest group. In the cross-domain transfer phase, we incorporate an interest meta-network module to create personalized interest bridges for effective preference transfer. Additionally, we enhance the model with a semi-supervised learning strategy that leverages pseudo-overlapping user data to mitigate data sparsity. Consequently, comprehensive experiments confirm that our model surpasses existing state-of-the-art methods.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128638"},"PeriodicalIF":7.5,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144331244","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"STLLM-GAN: Spatio-temporal LLM Generative Adversarial Network for PM2.5 prediction","authors":"Changkui Yin ,&nbsp;Yingchi Mao ,&nbsp;Liren Deng ,&nbsp;Meng Chen ,&nbsp;Yi Rong ,&nbsp;Xiaoming He ,&nbsp;Xiaofeng Zhou","doi":"10.1016/j.eswa.2025.128250","DOIUrl":"10.1016/j.eswa.2025.128250","url":null,"abstract":"<div><div>Accurate <span><math><mrow><mi>P</mi><msub><mi>M</mi><mrow><mn>2.5</mn></mrow></msub></mrow></math></span> prediction plays an important role in climate change mitigation and environmental protection. As an emerging artificial intelligence technique, Large Language Models (LLMs) have exhibited powerful data processing and adaptive feature learning capabilities, thus being widely applied in Time Series Prediction (TSP). Unfortunately, the current LLM-based TSP models pose difficulties in accurate <span><math><mrow><mi>P</mi><msub><mi>M</mi><mrow><mn>2.5</mn></mrow></msub></mrow></math></span> prediction due to two factors: 1) they neglect or fail to fully extract spatio-temporal dependencies, and 2) these LLM-based TSP models solely rely on the supervised training, failing to learn real data’s distribution. Jointly considering these factors, in this paper, we present a novel <span><math><mrow><mi>P</mi><msub><mi>M</mi><mrow><mn>2.5</mn></mrow></msub></mrow></math></span> prediction framework namely Spatio-Temporal LLM Generative Adversarial Network (STLLM-GAN). In detail, to capture spatio-temporal dependencies, Spatio-Temporal Large Language Model (STLLM) is first developed, containing a Spatio-Temporal Module (STM) and an LLM-enabled Inference Module (LLMIM). For the purpose of optimizing STLLM’s training, we design an adversarial training scheme using Generative Adversarial Network. The scheme incorporates un- and supervised training through a jointly beneficial manner. The unsupervised training seeks to learn real data’s distribution, while the supervised training strives to align the actual values with estimations by minimizing Mean Squared Error (MSE) function. We carry out extensive experiments on two real-world air pollutant concentration datasets, covering Shanghai city and Beijing city, respectively. The experimental results prove that STLLM-GAN is superior to advanced benchmarks in prediction performance.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128250"},"PeriodicalIF":7.5,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144331245","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A review of the frameworks, models, and algorithms for large-scale imaging satellite mission planning","authors":"Xiutian Li ,&nbsp;Yingwu Chen ,&nbsp;Lining Xing ,&nbsp;Yingguo Chen ,&nbsp;Yonghao Du ,&nbsp;Lei He","doi":"10.1016/j.eswa.2025.128471","DOIUrl":"10.1016/j.eswa.2025.128471","url":null,"abstract":"<div><div>Imaging satellite mission planning plays a central and crucial role in the operation control of large-scale imaging satellite constellations. To such satellite constellations, the mission planning frameworks and models for single satellites and small-scale satellite constellations are not applicable any longer, while some main algorithms can still be used. In such a context, the frameworks, models, and algorithms for large-scale imaging satellite mission planning (LSISMP) are reviewed, and some possible future research directions are pointed out. Firstly, three types of planning frameworks (centralized, distributed, and centralized-distributed ones) are discussed, and their advantages and disadvantages as well as application scenarios are deeply analyzed. The importance and role of different mission planning frameworks in large-scale imaging satellites are revealed. Then, the decision forms and common features of LSISMP are unveiled from perspectives of mathematical programming models, combinatorial optimization models, and so on. Based on the idea of increasing intelligence, the algorithms are teased from three dimensions, namely, from the conventional exact algorithms, to metaheuristic algorithms, and finally to highly intelligent machine learning algorithms. Finally, the research proposes that LSISMP is expected to develop towards the autonomous modeling and the deep integration of machine learning and intelligent optimization algorithms.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128471"},"PeriodicalIF":7.5,"publicationDate":"2025-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322783","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-feature representation-based graph attention networks for predicting potential supply relationships in a large-scale supply chain network","authors":"Donghun Lee ,&nbsp;Jimin Go ,&nbsp;Taehyun Noh ,&nbsp;Seokwoo Song","doi":"10.1016/j.eswa.2025.128593","DOIUrl":"10.1016/j.eswa.2025.128593","url":null,"abstract":"<div><div>This study aims to predict potential supply relationships within a large-scale supply chain network. Identifying appropriate suppliers can help companies mitigate disruptions in the supply of materials and finances. Furthermore, it offers the companies potential profits by enabling more effective resource allocation and fostering innovation. While previous studies have adopted machine learning approaches, these methods may not fully capture the complexity of network topology. Graph neural network-based methods have recently gained attention as a promising alternative. However, since graph neural network-based methods mainly rely on fixed aggregation weights, these methods often struggle to capture the complexity of supply relationships between companies. This study proposes multi-feature representation-based graph attention networks, which explore hidden topological relationships between companies by incorporating semantic characteristics such as product and network features. Our findings demonstrate that the proposed method outperforms machine learning-based and state-of-the-art graph neural network-based methods. In addition, ablation studies confirm that the proposed components significantly improve prediction performance.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128593"},"PeriodicalIF":7.5,"publicationDate":"2025-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322784","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A laser pointer-based interaction method for providing robot assistance to people with upper limb impairments","authors":"Yan Liu ,&nbsp;Yaxin Liu ,&nbsp;Yufeng Yao,&nbsp;Ming Zhong","doi":"10.1016/j.eswa.2025.128562","DOIUrl":"10.1016/j.eswa.2025.128562","url":null,"abstract":"<div><div>As the global population ages, people with upper limb impairments are increasing. Limited healthcare resources struggle to meet the growing nursing demand. Recently, a laser-guided robotic assistance technology has been proposed. Characterized by low cost, intuitive operation, and user-friendly features, it has shown great potential in enhancing the independence and daily participation of people with disabilities. However, most research on laser pointer-guided robot assistance has two challenges, one is the interaction is often disrupted due to the noise spots in reflective and refractive environments, and the other is the predefined action execution under the laser guidance is difficult to reasonably accomplish the task. To address these problems, this paper presents a laser pointer-based interaction system for robotic assistance. First, a novel laser pointer with an ArUco marker is developed, which contributes to filtering the noise spots and detecting the intent spot. Then, a shared autonomy mechanism for laser pointer-based interaction is introduced, which can not only utilize laser spots to prompt the robot to perform predefined tasks but also guide the robot to perform non-predefined tasks during the execution. Finally, the proposed method is tested on the six reflective and refractive environments and four household tasks. Six volunteers are recruited to perform a subjective evaluation. Results demonstrate the effectiveness of the proposed system and its feasibility in assisting individuals with upper limb impairments.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128562"},"PeriodicalIF":7.5,"publicationDate":"2025-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322788","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A cumulative capital approach for dynamic transmission expansion planning: enhancing cost efficiency and grid development","authors":"Salman Habib","doi":"10.1016/j.eswa.2025.128665","DOIUrl":"10.1016/j.eswa.2025.128665","url":null,"abstract":"<div><div>This study introduces a novel cumulative capital approach for dynamic transmission expansion planning (DTEP), enabling planners to carry over unspent budget across multiple years. Traditional models with rigid annual budgets often lead to investment infeasibility and suboptimal infrastructure development. In contrast, the proposed mixed-integer linear programming (MILP) model integrates budget carryover constraints, expanding the feasible solution space and enabling more strategic long-term investments. Simulation results on 6-bus, 24-bus, and 118-bus networks show that the proposed model achieves substantial improvements: load shedding is reduced by up to 88%, total costs by up to 53%, and the model maintains feasibility under tight budgets where classical models fail. For example, in budget-constrained scenarios, critical transmission lines that are unaffordable under annual caps become viable when capital is accumulated, allowing early resolution of network congestion. Furthermore, the model exhibits consistent scalability and robustness, with computation times acceptable for practical use even on large networks. These findings establish the cumulative capital approach as a cost-effective and technically sound strategy for transmission infrastructure planning under fiscal constraints. This work provides a valuable tool for policymakers and system planners aiming to balance reliability, cost, and regulatory flexibility over multi-year horizons.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128665"},"PeriodicalIF":7.5,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144331234","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Scaffold-driven molecular generation via reinforced RNN with centroid distance evaluation","authors":"Xingzheng Zhu ,&nbsp;Zhihong Zhao ,&nbsp;Fei Zhu","doi":"10.1016/j.eswa.2025.128606","DOIUrl":"10.1016/j.eswa.2025.128606","url":null,"abstract":"<div><div>De novo molecular design is learning from existing data to propose a new chemical structure that meets the desired properties. Still, it is difficult to de novo design a variety of novel molecules with desirable properties because the different properties of molecules cannot be balanced using a simple generative method. To solve this problem, this study proposes a reinforced Pareto-optimized molecular scaffold clustering generation method, ScaRL-P. Molecular scaffold information can help us identify molecules of the same pattern, cluster according to the core characteristics of the scaffold, and screen out the molecules with ideal properties. In addition, this study uses Pareto optimization to construct a three-dimensional Pareto frontier - biological activity, diversity, and in-cluster reward value, and cooperate with molecular scaffold clustering to obtain the dominant molecules. The multi-dimensional frontier obtained by adjusting the Pareto frontier in reinforcement learning modeling is transformed into the final reward return, which is provided to the agent in reinforcement learning to learn a molecular generation strategy that is close to the optimal attribute distribution.ScaRL-P demonstrated superior performance in binding affinity and optimization for three protein objectives (KOR, PIK3CA, and JAK2), outperforming several GPC benchmark methods.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128606"},"PeriodicalIF":7.5,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322779","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-objective collaborative path planning for multiple water-air unmanned vehicles in cramped environments","authors":"Shihong Yin ,&nbsp;Jiabao Hu ,&nbsp;Zhengrong Xiang","doi":"10.1016/j.eswa.2025.128625","DOIUrl":"10.1016/j.eswa.2025.128625","url":null,"abstract":"<div><div>Water-air amphibious unmanned vehicle (WAAUV) systems are highly adaptable to complex and confined workspaces, offering tremendous potential for tasks such as search and rescue. However, planning safe and efficient cooperative paths for multiple WAAUVs in crowded environments remains challenging due to trajectory conflicts associated with high-density navigation and increased collision risks under space constraints. This paper proposes an improved multitasking-constrained multi-objective optimization (IMTCMO) for the collaborative path planning problem. The algorithm employs a multitasking coevolutionary framework with dynamic constraint relaxation and hybrid differential evolution operators. It optimizes main and auxiliary tasks simultaneously, balancing global exploration and local exploitation. A multi-vortex superposition model is employed to quantify environmental disturbances, and a model for WAAUV path planning is constructed, incorporating objectives for task collaboration efficiency, threat risk cost, and energy consumption. In addition, an adaptive coding strategy is designed to improve solution quality. Experiments in six complex scenarios show that IMTCMO outperforms seven advanced algorithms in convergence, diversity, and robustness, improving average hypervolume by 1.59 %. Even in multi-threat areas with complex fluid dynamic interference, IMTCMO can still generate efficient, safe, and low-energy cooperative paths.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128625"},"PeriodicalIF":7.5,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144314215","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Joint optimization of spare parts inventory and maintenance for wind turbine systems","authors":"Haibo Jin ,&nbsp;Huawei Li ,&nbsp;Jiayu Bi ,&nbsp;Mengjiao Li","doi":"10.1016/j.eswa.2025.128588","DOIUrl":"10.1016/j.eswa.2025.128588","url":null,"abstract":"<div><div>This paper focuses on optimizing maintenance and spare parts inventory strategies for wind turbines, specifically targeting a four-component system with series-parallel structural. We explore the joint decision-making and optimization problems of maintenance and spare parts inventory by modeling the deterioration process of the system using stochastic processes. Moreover, we formulate corresponding maintenance plans and spare parts ordering strategies for components with different degrees of importance by considering the structural and stochastic correlations among components, as well as the delay in spare parts supply. To address the challenges of joint optimization, we develop a hybrid algorithm that combines Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), termed the GA-PSO algorithm, to solve the strategies of the constructed model. Finally, it is verified that the effectiveness, high performance and robustness, of the proposed strategies, algorithms, through numerical case analysis, comparative experiments as well as analysis regarding the impact of key parameters on the model. Sensitivity analysis reveals that random correlation between key and non-key components has relatively high sensitivity on the model, which should be paid more attention.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128588"},"PeriodicalIF":7.5,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144321520","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep learning framework for enhanced MRI analysis in healthcare diagnosis","authors":"Xue Wen","doi":"10.1016/j.eswa.2025.128487","DOIUrl":"10.1016/j.eswa.2025.128487","url":null,"abstract":"<div><div>Artificial intelligence for computer vision has improved a lot due to deep learning, but making it work better is still a big task. This is particularly important for demanding areas like diagnostic imaging that need high levels of precision and effectiveness. This study aims to improve deep learning methods to make them better for use in healthcare diagnosis, especially in analyzing MRIs. Traditional methods like Least Squares Regression,<!--> <!-->Random Forests, CNN, and<!--> <!-->Support Vector Machine<!--> <!-->can have difficulty managing big datasets, adapting to different situations, and being efficient in their calculations. In order to overcome these constraints, the suggested system adheres to a systematic procedure: In collecting data: We use a detailed MRI dataset like the BraTS (Brain Tumor Segmentation) dataset to gather a variety of labeled medical pictures for effective training. In<!--> <!-->initial processing: denoising methods are used to improve the appearance of MRI images and<!--> <!-->are standardized to a uniform scale; In<!--> <!-->augmenting data: basic augmentation methods like flipping, rotating, and increasing the intensity are used to increase the volume of data and the system’s generalizability; In training the model with transfer learning: The EfficientNet-B4 system is used over<!--> <!-->the MRI dataset. This<!--> <!-->model’s structure is effective and can easily scale, allowing it to pick up important traits for diagnostic imaging. In<!--> <!-->optimizing<!--> <!-->the model’s performance:<!--> <!-->Bayesian optimization<!--> <!-->method is used to tweak hyperparameters, making sure the model is configured optimally to maximize precision while reducing the consumption of resources. Factors such as accuracy, precision, recall, computational efficiency, and robustness<!--> <!-->are used to thoroughly assess the framework. Using these new methods, our system successfully tackles problems in medical imaging and improves DL<!--> <!-->in computer vision technology. This leads to better and more efficient tools for diagnosing health issues.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128487"},"PeriodicalIF":7.5,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144313972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}