Expert Systems with Applications最新文献

{"title":"Comprehensive analysis of Transformer networks in identifying informative sentences containing customer needs","authors":"Mehrshad Kashi ,&nbsp;Salim Lahmiri ,&nbsp;Otmane Ait Mohamed","doi":"10.1016/j.eswa.2025.126785","DOIUrl":"10.1016/j.eswa.2025.126785","url":null,"abstract":"<div><div>The unprecedented rise in user-generated content (UGC) provides businesses with new opportunities to extract customer insights from unstructured data, particularly for identifying customer needs. Intelligent methods offer time- and cost-efficient solutions to extract such insights from the plethora of repetitive and often redundant UGC. However, widespread adoption of these methods faces significant barriers, including high deployment and maintenance costs, data availability challenges, task complexity, and concerns about model efficacy and ethical implications. To facilitate broader adoption of intelligent systems in customer needs analysis, this study evaluates Transformer-based models in terms of generalizability, fairness, robustness, and sample efficiency across various experimental settings to uncover their true performance and identify the root causes of their errors. Our results show that although Transformer-based models improved the F1-score by up to 18% compared to baselines, their limitations become evident when evaluating their performance against task objectives. Key findings include: (i) Transformer-based networks share error patterns and struggle to identify infrequent or unseen informative samples, (ii) they heavily rely on abundant information and lexical cues for accurate predictions, compromising inter- and intra-domain generalizability, (iii) larger models do not necessarily improve sample efficiency within their domain, and (iv) while optimal cross-domain results arise from complex domain training, adding more in-domain samples does not enhance cross-domain performance. Overall, this research provides crucial insights to help businesses overcome adoption barriers when implementing Natural Language Processing advancements, such as Transformer-based models, in the customer needs analysis process. Source codes are available at <span><span>https://github.com/mehrshad-kashi/ISCN-UsingTransformerNetworks</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126785"},"PeriodicalIF":7.5,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143428891","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":"EfficientFaceV2S: A lightweight model and a benchmarking approach for drone-captured face recognition","authors":"Mohamad Alansari ,&nbsp;Khaled Alnuaimi ,&nbsp;Iyyakutti Ganapathi ,&nbsp;Sara Alansari ,&nbsp;Sajid Javed ,&nbsp;Abdulhadi Shoufan ,&nbsp;Yahya Zweiri ,&nbsp;Naoufel Werghi","doi":"10.1016/j.eswa.2025.126786","DOIUrl":"10.1016/j.eswa.2025.126786","url":null,"abstract":"<div><div>Face recognition in aerial imagery encounters distinctive challenges, including low resolution and varying pitch angles. The influence of image quality, particularly resolution, on the performance of existing face recognition systems is well established. However, limited exploration exists regarding the performance of FR models on drone-captured images, primarily due to the scarcity of suitable datasets. To address this gap, our study investigates the efficacy of face recognition models when applied to drone-captured facial images. We utilize state-of-the-art lightweight and large models, evaluating them across three drone-captured benchmarks and one dataset focused on low resolution. To ensure a comprehensive evaluation, we additionally adopt seven widely recognized benchmarks, which are artificially downsampled and rotated to simulate the impact of distance and altitude on the view from a drone to a target. Our results highlight a substantial decrease in accuracy across all FR models in these challenging scenarios. In response to this challenge, we introduce a model, EfficientFaceV2S. The proposed EfficientFaceV2S model demonstrates consistent performance across all benchmarks while imposing minimal computational demands. This makes it particularly suitable for real-time and resource-constrained applications. The significance of our work lies in the development of EfficientFaceV2S, which effectively addresses the unique challenges posed by drone-captured images, offering significant improvements in accuracy and efficiency over existing models.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126786"},"PeriodicalIF":7.5,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143420569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prior knowledge evaluation and emphasis sampling-based evolutionary algorithm for high-dimensional medical data feature selection","authors":"Zhilin Wang ,&nbsp;Lizhi Shao ,&nbsp;Ali Asghar Heidari ,&nbsp;Mingjing Wang ,&nbsp;Huiling Chen","doi":"10.1016/j.eswa.2025.126737","DOIUrl":"10.1016/j.eswa.2025.126737","url":null,"abstract":"<div><div>Handling high-dimensional medical data presents a significant challenge. Numerous irrelevant and redundant features impede the construction of high-precision models. Traditional Feature Selection (FS) algorithms often fail to address the nonlinear and combinatorial relationships inherent in high-dimensional medical features. To address these issues, we propose the Prior Knowledge Evaluation Strategy (PES) and Emphasis Sampling Strategy (ESS) based Harris Hawks Optimization (PSHHO) algorithm. The PES strategy stores the historical optimal solutions in an archive, effectively eliminating low-quality and invalid solutions, thereby reducing evaluation time. The ESS mechanism enables the Harris Hawk Optimization algorithm to fully utilize the valuable information embedded in the optimal solutions, enhancing its optimization performance through equidistant sampling for local exploitation. Additionally, a V-shaped transfer function is employed to convert the algorithm into its binary version, BPSHHO, for FS in high-dimensional medical data. The experimental results demonstrate that the PSHHO algorithm significantly outperforms the compared algorithms on the CEC benchmark test set, achieving the global best results on 17 out of 30 test functions. It exhibits excellent performance on medical datasets with dimensions exceeding 5000. Compared to other binary algorithms, BPSHHO achieves the highest classification accuracy with the fewest features. On 9 high-dimensional medical datasets, it achieved top accuracy in 8 cases using classifiers constructed with no more than 15 features. This highlights its effectiveness as a FS method for high-dimensional medical data.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126737"},"PeriodicalIF":7.5,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143438162","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":"Stochastic unit commitment problem: A statistical approach","authors":"Carlos Olivos ,&nbsp;Jorge Valenzuela","doi":"10.1016/j.eswa.2025.126787","DOIUrl":"10.1016/j.eswa.2025.126787","url":null,"abstract":"<div><div>The Stochastic Unit Commitment Problem (SUCP) has been widely studied using scenario-based generation to include uncertainty in the mathematical model, transforming the stochastic problem into a large deterministic problem. However, the accuracy of the stochastic problem is highly dependent on the number of scenarios, leading to computational intractability when the number of scenarios is large. This paper proposes a novel paradigm that avoids scenario sampling. Instead, it derives a function that models the expected cost based on a merit order dispatch rule for the thermal units and incorporates the probability distribution of net demand. Thus, the expected cost is explicitly stated in a non-linear function. A piecewise linear approximation method is used to address the new model’s nonlinearity, resulting in a mixed integer linear programming (MILP) model. The proposed model is compared to the traditional scenario-based SUCP in terms of computational effort, solution stability, and costs. Numerical experiments show that the new approach can reach optimality in more instances than the traditional scenario-based model. Moreover, it eliminates memory limitations and provides stable and cost-competitive solutions. Thus resulting in a scalable alternative for large-scale and realistic power systems. To the best of our knowledge, this is the first SUCP formulation that integrates uncertainty without relying on scenario-based methods.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126787"},"PeriodicalIF":7.5,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143403067","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":"Neural Chronos ODE: Modelling bidirectional temporal patterns in time-series data","authors":"C. Coelho ,&nbsp;M. Fernanda P. Costa ,&nbsp;L.L. Ferrás","doi":"10.1016/j.eswa.2025.126784","DOIUrl":"10.1016/j.eswa.2025.126784","url":null,"abstract":"<div><div>This work introduces Neural Chronos Ordinary Differential Equations (Neural CODE), a deep neural network architecture that fits a continuous-time ODE dynamics for predicting the chronology of a system both forward and backward in time. To train the model, we solve the ODE as an initial value problem and a final value problem, similar to Neural ODEs. We also explore two approaches to combining Neural CODE with Recurrent Neural Networks by replacing Neural ODE with Neural CODE (CODE-RNN), and incorporating a bidirectional RNN for full information flow in both time directions (CODE-BiRNN). Variants with other update cells namely GRU and LSTM are also considered and referred to as: CODE-GRU, CODE-BiGRU, CODE-LSTM, CODE-BiLSTM. Experimental results demonstrate that Neural CODE outperforms Neural ODE in learning the dynamics of a spiral forward and backward in time, even with sparser data. We also compare the performance of CODE-RNN/-GRU/-LSTM and CODE-BiRNN/-BiGRU/-BiLSTM against ODE-RNN/-GRU/-LSTM on three real-life time-series data tasks: imputation of missing data for lower and higher dimensional data, and forward and backward extrapolation with shorter and longer time horizons. Our findings show that the proposed architectures converge faster, with CODE-BiRNN/-BiGRU/-BiLSTM consistently outperforming the other architectures on all tasks, achieving a notably smaller mean squared error—often reduced by up to an order of magnitude.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126784"},"PeriodicalIF":7.5,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143429276","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":"Pedestrian dynamics model for high densities","authors":"Grzegorz Bazior,&nbsp;Jarosław Wąs,&nbsp;Dariusz Pałka","doi":"10.1016/j.eswa.2025.126775","DOIUrl":"10.1016/j.eswa.2025.126775","url":null,"abstract":"<div><div>This paper presents a novel Cellular Automata (CA) model for crowd simulations, aimed at understanding pedestrian dynamics in high-density environments. The model incorporates static and dynamic floor fields, multi-cell representations of pedestrians, and a mechanism for simulating blockades, where individuals may be pushed towards walls. Inspired by Burstedde’s work, our model enhances realism and accuracy in crowd behavior simulations. It was validated against real-world experiments involving 68 students, focusing on pedestrian density, cumulative percent flow through bottlenecks, and Voronoi density. Comparisons with the Burstedde et al. model show that our approach provides more accurate simulations for scenarios involving highly motivated pedestrians navigating through bottlenecks. These findings contribute to the development of more effective crowd management strategies and advanced simulation techniques.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"272 ","pages":"Article 126775"},"PeriodicalIF":7.5,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143394529","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SC4ANM: Identifying optimal section combinations for automated novelty prediction in academic papers","authors":"Wenqing Wu,&nbsp;Chengzhi Zhang,&nbsp;Tong Bao,&nbsp;Yi Zhao","doi":"10.1016/j.eswa.2025.126778","DOIUrl":"10.1016/j.eswa.2025.126778","url":null,"abstract":"<div><div>Novelty is a core component of academic papers, and there are multiple perspectives on the assessment of novelty. Existing methods often focus on word or entity combinations, which provide limited insights. The content related to a paper’s novelty is typically distributed across different core sections, e.g., Introduction, Methodology and Results. Therefore, exploring the optimal combination of sections for evaluating the novelty of a paper is important for advancing automated novelty assessment. In this paper, we utilize different combinations of sections from academic papers as inputs to drive language models to predict novelty scores. We then analyze the results to determine the optimal section combinations for novelty score prediction. We first employ natural language processing techniques to identify the sectional structure of academic papers, categorizing them into introduction, methods, results, and discussion (IMRaD). Subsequently, we used different combinations of these sections (e.g., introduction and methods) as inputs for pretrained language models (PLMs) and large language models (LLMs), employing novelty scores provided by human expert reviewers as ground truth labels to obtain prediction results. The results indicate that using introduction, results and discussion is most appropriate for assessing the novelty of a paper, while the use of the entire text does not yield significant results. Furthermore, based on the results of the PLMs and LLMs, the introduction and results appear to be the most important section for the task of novelty score prediction. The code and dataset for this paper can be accessed at <span><span>https://github.com/njust-winchy/SC4ANM</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126778"},"PeriodicalIF":7.5,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143403068","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":"The bi-objective mixed-fleet vehicle routing problem under decentralized collaboration and time-of-use prices","authors":"Weixuan Shi ,&nbsp;Nengmin Wang ,&nbsp;Li Zhou ,&nbsp;Zhengwen He","doi":"10.1016/j.eswa.2025.126875","DOIUrl":"10.1016/j.eswa.2025.126875","url":null,"abstract":"<div><div>Electric vehicles (EVs) can effectively reduce transportation carbon emissions. However, their limited driving range, longer charging times, and scarce charging locations make their transportation efficiency lower compared to traditional internal combustion engine vehicles (ICEVs). A mixed fleet leverages the strengths of both vehicle types. Additionally, collaborative logistics can further enhance these strengths by improving vehicle utilization. Therefore, this study proposes a mixed-fleet model within a collaborative logistics framework to enhance transportation efficiency and balance carbon emission reductions and economic benefits. Considering the variability in charging prices, we developed a bi-objective mixed-fleet vehicle routing optimization model with time windows, incorporating order selection and time-of-use electricity pricing. An ε-constraint clustering hybrid evolutionary algorithm is formulated based on the problem characteristics. Numerical experiments with standard and large-scale instances verified the efficiency and superior performance of the developed model and algorithm. Finally, a sensitivity analysis provided managerial insight.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126875"},"PeriodicalIF":7.5,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143420564","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":"An empirical study of best practices for code pre-trained models on software engineering classification tasks","authors":"Yu Zhao,&nbsp;Lina Gong,&nbsp;Yaoshen Yu,&nbsp;Zhiqiu Huang,&nbsp;Mingqiang Wei","doi":"10.1016/j.eswa.2025.126762","DOIUrl":"10.1016/j.eswa.2025.126762","url":null,"abstract":"<div><div>Tackling code-specific classification challenges like detecting code vulnerabilities and identifying code clones is pivotal in software engineering (SE) practice. The utilization of pre-trained models (PTMs) from the natural language processing (NLP) field shows profound benefits in text classification by generating contextual token embeddings. Similarly, for code-specific classification tasks, there is a growing trend among researchers and practitioners to leverage code-oriented PTMs to create embeddings for code snippets or directly apply the code PTMs to the downstream tasks based on the pre-training and fine-tuning paradigm. Nonetheless, we observe that SE researchers and practitioners often treat the code and text in the same way as NLP strategies when employing these code PTMs. However, despite previous studies in the SE field indicating similarities between programming languages and natural languages, it may not be entirely appropriate for current researchers to directly apply NLP knowledge to assume similar behavior in code. Therefore, in order to derive best practices for researchers and practitioners to use code PTMs for SE classification tasks, we first conduct an empirical analysis on six distinct code PTMs, namely CodeBERT, StarEncoder, CodeT5, PLBART, CodeGPT, and CodeGen, across three architectural frameworks (encoder-only, decoder-only, and encoder–decoder) in the context of four SE classification tasks: code vulnerability detection, code clone identification, just-in-time defect prediction, and function docstring mismatch detection under two scenarios of code embedding and task model. Our findings reveal several insights on the use of code PTMs for code-specific classification tasks endeavors: (1) Emphasizing the vector representation of individual code tokens leads to better code embedding quality and task model performance than those generated through specific tokens techniques in both the code embedding scenario and task model scenario. (2) Larger-sized code PTMs do not necessarily lead to superior code embedding quality in the code embedding scenario and better task performance in the task model scenario. (3) Adopting the ways to handle code and text data same as the pre-training phrase cannot guarantee the acquisition of high-quality code embeddings in the code embedding scenario while in the task model scenario, it can most likely acquire better task performance.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"272 ","pages":"Article 126762"},"PeriodicalIF":7.5,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143394525","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":"Intelligent vehicle decision-making strategy integrating spatiotemporal features at roundabout","authors":"Wenxiao Ma ,&nbsp;Jian Wu ,&nbsp;Bohua Sun ,&nbsp;Xinlun Leng ,&nbsp;Weiwei Miao ,&nbsp;Zhenhai Gao ,&nbsp;Wenjin Li","doi":"10.1016/j.eswa.2025.126779","DOIUrl":"10.1016/j.eswa.2025.126779","url":null,"abstract":"<div><div>To overcome the problems of weak driving safety and low traffic efficiency of intelligent vehicles in roundabout scenarios, and to improve the autonomous decision-making ability of intelligent systems. In this paper, we propose an intelligent vehicle Decision-Making Strategy based on SpatioTemporal graph neural Networks, namely DMS-STNet. Using end-to-end deep learning methods based on the historical information of intelligent vehicles and surrounding vehicles, output the action sequence of future driving behavior of intelligent vehicles. Specifically, a spatiotemporal graph is used to model the driving environment of vehicles, and a graph convolutional neural network is used to explore the spatial interaction relationship between intelligent vehicles and environmental vehicles. Next, based on the time convolutional network, learn the temporal characteristics of intelligent vehicles. Further integrate the complex spatiotemporal interaction relationship between intelligent vehicles and environmental vehicles through a gated fusion network. Moreover, a multi-layer perceptron is used to map the fused tensor into a sequence of driving behavior actions. In addition, experimental data collection and software in the loop testing verification were conducted on the Carla simulator platform. The research results indicate that the model proposed in this paper outperforms the comparative models in terms of prediction accuracy, safety, and traffic efficiency, fully leveraging the autonomous decision-making performance advantages of intelligent vehicles.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126779"},"PeriodicalIF":7.5,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143420469","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}