Expert Systems with Applications最新文献

{"title":"Insights into the impact of visual and textual information on investment decision-making: A multimodal business plan analysis via deep representation learning","authors":"Weikang Yuan ,&nbsp;Tianqianin Lin ,&nbsp;Zhuoren Jiang ,&nbsp;Song Wang","doi":"10.1016/j.eswa.2025.128911","DOIUrl":"10.1016/j.eswa.2025.128911","url":null,"abstract":"<div><div>Business plans (BPs) serve as crucial communication tools between entrepreneurs and investors, but there is controversy in existing research regarding the actual impact of BP’s quality on investment decision. To investigate how the visual and textual features of BP impact investors’ investment decisions, we develop a flexible computational framework to represent the visual and textual information in business plans and design a series of indicators to measure the corresponding information quality. Specifically, we propose three quality indicators, namely <span><math><msub><mi>V</mi><mrow><mi>B</mi><mspace></mspace><mi>P</mi></mrow></msub></math></span>, <span><math><msub><mi>T</mi><mrow><mi>B</mi><mspace></mspace><mi>P</mi></mrow></msub></math></span>, and <span><math><msub><mi>I</mi><mrow><mi>B</mi><mspace></mspace><mi>P</mi></mrow></msub></math></span>, based on the deep representations of the BP’s visual feature, text information, and brief introduction. Through Logit Regression analysis of 4597 business plans and their corresponding 42,533 decision-making samples from an online investment platform, we find BP’s quality significantly influences initial investment decisions. Visual quality and textual introduction quality exhibit significant effects (p &lt; 0.05). We also reveal the moderating effects of investor risk preferences. Our computational modeling and empirical evidence provide key insight into decision mechanisms. This is the first investigation that utilizes deep pre-trained models to comprehensively model BP’s multimodal features within the investment decision-making domain. The proposed quality indicators can enable scalable, unbiased evaluation to address the evolving needs of decision-makers in an increasingly complex and data-rich environment.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128911"},"PeriodicalIF":7.5,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144605719","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":"Content suppression mechanisms-based recommendation systems","authors":"Haifeng Yang ,&nbsp;Ran Zhang ,&nbsp;Jianghui Cai ,&nbsp;Jie Wang ,&nbsp;Yupeng Wang ,&nbsp;Yating Li ,&nbsp;Yaling Xun ,&nbsp;Xujun Zhao","doi":"10.1016/j.eswa.2025.128928","DOIUrl":"10.1016/j.eswa.2025.128928","url":null,"abstract":"<div><div>Personalized recommendation systems enhance user experiences but often lead to information filter bubbles and reduced content diversity. To address these issues, this paper introduces a novel recommendation system based on Content Suppression Mechanisms, centered around Restrain Gated Recurrent Units (RGRU). The core innovation lies in a suppression function that dynamically adjusts the likelihood of recommending items based on their similarity to previously viewed content. This approach effectively mitigates the repetition of similar items andpromotes diversity within recommendation lists. Furthermore, we introduce a novel evaluation metric, the External and Intra-List Similarity (<span><math><mrow><mi>E</mi><mo>&amp;</mo><mi>I</mi><mi>L</mi><mi>S</mi></mrow></math></span>), designed to assess both the internal diversity of recommended items and their deviation from previous interactions of users. This metric improves upon existing diversity metrics by addressing systemic diversity and variations among user groups.Validation across KuaiRec, ml_25m, and MRM datasets demonstrates that our approach maintains high precision while significantly enhancing recommendation diversity. This dual improvement facilitates multi-perspective content exploration, mitigating information cocoons and elevating user experience.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128928"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144631519","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 similarity-based semi-supervised algorithm for labeling unlabeled text data","authors":"Kirankumar Singh Potshangbam,&nbsp;Kshetrimayum Nareshkumar Singh","doi":"10.1016/j.eswa.2025.128941","DOIUrl":"10.1016/j.eswa.2025.128941","url":null,"abstract":"<div><div>This paper presents a novel, non-iterative semi-supervised learning algorithm that leverages cosine similarity between document vectors and class mean vectors to label unlabeled text data automatically. The proposed method supports multiple vectorization techniques, including CountVectorizer, TF-IDF, and Doc2Vec, and is classifier-agnostic, enabling compatibility with both traditional and deep learning models such as KNN, Multinomial Naïve Bayes, SGDClassifier, Logistic Regression, Feedforward Neural Networks (FNN), and Convolutional Neural Networks (CNN). Extensive experiments conducted on benchmark datasets (BBC, Inshorts, 20-newsgroups) demonstrate: (1) achieving 96.88% accuracy on BBC, 93.59% on Inshorts, and 92.49% on 20-newsgroups with only 30% labeled data, thereby reducing manual labeling effort by over 99%; (2) TF-IDF consistently outperforms CountVectorizer and Doc2Vec by 3–12 percentages in accuracy across most experimental settings; and (3) Logistic Regression and FNN achieve the best performance among the classifiers. The method offers a practical, resource-efficient solution for real-world text classification by bridging labeled-unlabeled data gaps without iterative retraining.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128941"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144613856","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":"Beyond aversion – principles of appropriate algorithmic decision-making in human resource management","authors":"Stefan Strohmeier,&nbsp;Mathias Becker,&nbsp;Ellen Scheer-Weller","doi":"10.1016/j.eswa.2025.128954","DOIUrl":"10.1016/j.eswa.2025.128954","url":null,"abstract":"<div><div>As algorithmic decision-making (ADM) becomes increasingly embedded in human resource management (HRM), concerns such as a lack of fairness and accountability raise urgent questions about its appropriateness. This study addresses the need for ADM evaluation by developing a coherent framework of principles grounded in the task-technology fit approach. It elaborates a balanced triad of nine indispensable ADM principles—methodical (veracity, accuracy, validity), managerial (relevancy, quality, efficiency), and ethical (fairness, accountability, transparency)—and validates them through a systematic literature review of 126 ADM artifacts in HRM. The analysis reveals a troubling lack of attention to ethical and managerial dimensions, while even methodical aspects are often neglected—with the notable exception of accuracy. Building on these findings, the study outlines a forward-looking agenda to operationalize, calibrate, implement, evaluate, and codify ADM principles, ultimately promoting responsible, appropriate ADM in HRM that reflects an evaluative stance beyond mere aversion.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128954"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633360","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":"iEBAKER: Improved remote sensing image-text retrieval framework via eliminate before align and keyword explicit reasoning","authors":"Yan Zhang,&nbsp;Zhong Ji,&nbsp;Changxu Meng,&nbsp;Yanwei Pang","doi":"10.1016/j.eswa.2025.128968","DOIUrl":"10.1016/j.eswa.2025.128968","url":null,"abstract":"<div><div>Recent studies focus on the Remote Sensing Image-Text Retrieval (RSITR), which aims at searching for the corresponding targets based on the given query. Among these efforts, the application of Foundation Models (FMs), such as CLIP, to the domain of remote sensing has yielded encouraging outcomes. However, existing FM based methodologies neglect the negative impact of weakly correlated sample pairs and fail to account for the key distinctions among remote sensing texts, leading to biased and superficial exploration of sample pairs. To address these challenges, we propose an approach named iEBAKER (an Improved Eliminate Before Align strategy with Keyword Explicit Reasoning framework) for RSITR. Specifically, we propose an innovative Eliminate Before Align (EBA) strategy to filter out the weakly correlated sample pairs, thereby mitigating their deviations from optimal embedding space during alignment.Further, two specific schemes are introduced from the perspective of whether local similarity and global similarity affect each other. On this basis, we introduce an alternative Sort After Reversed Retrieval (SAR) strategy, aims at optimizing the similarity matrix via reverse retrieval. Additionally, we incorporate a Keyword Explicit Reasoning (KER) module to facilitate the beneficial impact of subtle key concept distinctions. Without bells and whistles, our approach enables a direct transition from FM to RSITR task, eliminating the need for additional pretraining on remote sensing data. Extensive experiments conducted on three popular benchmark datasets demonstrate that our proposed iEBAKER method surpasses the state-of-the-art models while requiring less training data. Our source code will be released at https://github.com/zhangy0822/iEBAKER.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128968"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633482","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":"AIoT image analysis for real-time dispatching of shipyard transport devices: A focus on trailers","authors":"Youngjun Choo ,&nbsp;Sunghoon Lim ,&nbsp;Yonghyun Kim ,&nbsp;Yeojoon Park ,&nbsp;Yonghoon Oh ,&nbsp;Changyob Lee ,&nbsp;Wonjun Yun ,&nbsp;Namhun Kim","doi":"10.1016/j.eswa.2025.128947","DOIUrl":"10.1016/j.eswa.2025.128947","url":null,"abstract":"<div><div>In the shipbuilding industry, the Future of Shipyard (FOS) represents a new paradigm driven by data collection, analysis, and prediction. Among various logistics operations, trailer dispatching remains inefficient due to schedules being fixed days or even weeks in advance. To address this issue, we propose an AIoT-based wireless system that enables real-time trailer status monitoring by transmitting image data, GPS information, edge device status, and AI inference results. The proposed system integrates edge computers, wireless communication, centralized servers, and a deep learning model tailored for binary classification. The trailer’s complex operational status is decomposed into two tasks: (1) detecting location and movement via GPS, and (2) classifying loading status of ship components using image analysis. To classify the loading status from images, we evaluated five deep learning models-ResNet, VGG, EfficientNet, ViT, and VAN-based on accuracy and F1 score. Among them, the VGG model achieved the best performance, with 97.35 % accuracy and a 96.9 % F1 score, demonstrating its suitability for real-world deployment. To enhance model robustness in harsh industrial environments and varying device installation positions, we applied geometric augmentation and validated its effectiveness through additional experiments. Based on this AIoT wireless system, we introduce an innovative dispatch process that replaces manual, experience-based decision-making with data-driven intelligence. This leads to reductions in labor costs and process time, offering meaningful improvements for shipyard operations. The proposed framework also serves as a scalable reference for AIoT applications across broader industrial domains.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128947"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588204","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 generalized neural solver based on LLM-guided heuristic evoluation framework for solving diverse variants of vehicle routing problems","authors":"Minyan Chi ,&nbsp;Wei Pang ,&nbsp;Xuan Wu ,&nbsp;Peng Zhao ,&nbsp;YuanShu Li ,&nbsp;Tianfang Wang ,&nbsp;Junjie Qian ,&nbsp;Yubin Xiao ,&nbsp;Liupu Wang ,&nbsp;You Zhou","doi":"10.1016/j.eswa.2025.128876","DOIUrl":"10.1016/j.eswa.2025.128876","url":null,"abstract":"<div><div>Vehicle Routing Problems (VRPs) are key combinatorial optimization challenges with broad applications in logistics. While neural solvers based on attention mechanisms offer promising results, they require retraining for each VRP variant, limiting scalability. Existing expert-designed and LLM-based heuristic methods often suffer from limited exploration ability and premature convergence. We propose the Unified VRP Neural Solver (UNS), an LLM-enabled framework that dynamically adjusts attention scores by generating variant-specific heuristics without requiring retraining of neural model parameters. At its core, the LLM-Guided Heuristic Evolution (LHE) algorithm, which is inspired by population-based Differential Evolution (DE) frameworks, iteratively refines heuristics through Mutation, Global Crossover, and Local Crossover to enhance diversity and avoid local optima. Extensive experiments across 16 VRP variants show that LHE outperforms state-of-the-art neural solvers and LLM-based approaches. The similarity analysis of heuristic populations reveals that LHE maintains higher diversity and avoids premature convergence. Additional evaluations on CVRP and TSP, along with ablation studies, validate the effectiveness and generalizability of LHE.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128876"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588762","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":"Fresh-products community group-buying delivery problem for heterogeneous customers","authors":"Yankai Zhang ,&nbsp;Kaiqi Zhao ,&nbsp;Shiwei Liang ,&nbsp;Na Liu ,&nbsp;Shiyi Xu ,&nbsp;Bin Yu ,&nbsp;Wenxuan Shan","doi":"10.1016/j.eswa.2025.128984","DOIUrl":"10.1016/j.eswa.2025.128984","url":null,"abstract":"<div><div>Online community group-buying of fresh products has emerged as a popular model in urban e-commerce. This paper studies fresh products community group buying delivery problem of multiple commodities considering customer behavior. Compared with traditional fresh products e-commerce in which each customer is distributed individually, community group buying introduces a community leader to receive fresh products from distributors and residents in this community pick up their orders from this leader. The time gap between the distributor’s delivery to the community leader and the residents’ pick-up from the leader results in further deterioration of fresh products, which is the challenge in online community group-buying. We establish a distribution model considering deterioration of fresh products in refrigerated trucks and at the community leader’s location, in which three types of penalty costs are used to represent heterogeneous customer behaviors. Since different residents have separated delivery time windows, prioritizing delivery for which customers must be balanced. We design a memetic algorithm for this non-linear programming. A split algorithm considering multi-commodity delivery and time-varying arc costs is designed to improve the efficiency of memetic algorithm. Experiments show that the proposed method reduces total cost by an average of 13.18% compared to a commercial solver within a fixed time budget. The case study based on data from Beijing provides management insights. Specifically, delivery routes tend to prioritize communities with a higher concentration of retired residents, while increased customer diversity is associated with lower vehicle utilization rates.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128984"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144631570","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 intelligent wireless sensing algorithm for complex cross-domain scenarios based on DB-FA-YoLov6","authors":"Lingwei Xu ,&nbsp;Haiyang Sun ,&nbsp;Kai Wang ,&nbsp;Gaofeng Nie ,&nbsp;Zhe Chen ,&nbsp;T. Aaron Gulliver","doi":"10.1016/j.eswa.2025.128912","DOIUrl":"10.1016/j.eswa.2025.128912","url":null,"abstract":"<div><div>Wireless sensing technology can identify human motion via feature information from WiFi signals. The popularity of smartphones, wearable devices, and other smart devices has increased the use of wireless sensing in fields such as smart homes, smart healthcare, human–computer interaction, and autonomous vehicles. However, the mobile communication environment is complex and dynamic which makes wireless sensing challenging. The issues include low model sensing accuracy, poor scene generalization ability, and high environmental dependence. Therefore, this paper proposes a cross-domain intelligent wireless sensing algorithm based on a double branch frequency attention mechanism Yolov6 network called DB-FA-YoLov6. This integrates a Yolov6 neural network, frequency attention module, and residual module to provide efficient extraction of signal features and enhance model generalization. The goal is to reduce the effect of the environment on sensing tasks and improve robustness, portability, and cross-domain accuracy. The DB-FA-YOLOv6 model integrates two types of residual modules, BasicBlock and Bottleneck. It replaces the large modules in the Yolov6 network model with lightweight structures, which can decrease the number of parameters, improve the efficiency of model training and testing, and reduce the complexity. Compared with current sensing algorithms such as Vision Transformer Network for Multiple Vision Tasks (ViT-MVT), Environment Independent (EI), and Joint Adversarial Domain Adaptation (JADA), the proposed DB-FA-YOLOv6 algorithm has better sensing accuracy, sensing efficiency, and cross-domain performance. For the in-domain scenario, the proposed algorithm achieves improvements of 10.0 % in sensing accuracy and 10.1 % in sensing efficiency. The sensing accuracy of the proposed algorithm in cross-domain scenarios, namely location and orientation, is improved by 10.5 % and 9.7 %, and the sensing efficiency is improved by 7.0 % and 52.1 %, respectively.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128912"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588201","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":"Development and validation of interpretable machine learning models for photovoltaic panel temperature prediction","authors":"Bo Ren ,&nbsp;Qianggang Wang ,&nbsp;Niancheng Zhou ,&nbsp;Saad Mekhilef","doi":"10.1016/j.eswa.2025.128952","DOIUrl":"10.1016/j.eswa.2025.128952","url":null,"abstract":"<div><div>Accurate prediction of photovoltaic (PV) panel temperature is critical for optimizing the design, operation, and maintenance of PV systems. Although many steady-state and machine learning (ML) models have been proposed to characterize the relationship between meteorological elements and panel temperature, achieving a balance between prediction accuracy, interpretability, and extrapolation capability remains a challenge. Therefore, this study attempts to construct a PV panel temperature prediction framework that integrates feature engineering and interpretable ML techniques. A feature selection method combining Pearson correlation coefficient, Shapley additive explanations, and extreme gradient boosting quantitatively evaluates the correlation of meteorological elements and their contributions to temperature prediction. Furthermore, two symbolic regression methods based on genetic programming and multi-population evolutionary algorithms are employed to develop explicit models with concise expressions and excellent performance. Two experimental datasets are from a utility-scale PV plant and a commercial rooftop PV system, with sizes of 19383 × 6 and 4503 × 6, respectively. Experimental results show that the proposed method can accurately and reliably predict the operating temperature of different panels, achieving R² of 0.981 and 0.961. Comparative analyses highlight the superior accuracy, interpretability, and broad applicability of the proposed models. This work provides valuable insights for panel temperature prediction, interpretable ML model development, and PV system management.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"296 ","pages":"Article 128952"},"PeriodicalIF":7.5,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588203","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}