International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis The <strong>EIS</strong> journal is devoted to the publication of high quality papers in the field of intelligent systems applications in numerous disciplines. Original research papers, state-of-the-art reviews and technical notes are invited for publication. en-US <span>The submission of a paper implies that, if accepted for publication, it will not be published elsewhere in the same form, in any language, without the prior consent of the publisher. Before publication, authors are requested to assign copyright to CRL Publishing Ltd. This allows CRL to sanction photocopying, and to authorize the reprinting of issues or volumes according to demand. Authors' traditional rights will not be jeopardized by assigning Copyright in this way, as they retain the right to reuse the material following publication, and to veto third-party publication.</span> mydarshan.d@gmail.com (Darshan Dillon) ijcsse.tharam@gmail.com (Naeem Khalid Janjua) Sat, 07 Nov 2026 22:28:51 -0800 OJS 3.3.9.9 http://blogs.law.harvard.edu/tech/rss 60 A Study on the Utilization of Shared Bicycles in Comprehensive Passenger Transport Hub Area Based on Neural Network Evaluation https://website-eis.crlpublishing.com/index.php/eis/article/view/2048 <p>As a new mode of short-distance travel, shared bicycles play an important role in providing rapid first-/last-mile connections and have high potential for being utilized in the service areas of integrated passenger transport hubs. However, shared-bicycle use around such hubs still faces challenges, including relatively low utilization, uneven vehicle distribution, and irrational dispatching, which call for improvement and optimization. Based on the 2025 operational data of the public bicycle system in a high-speed railway hub area, this study investigates shared-bicycle utilization using big-data analytics and neural-network methods, and conducts analysis with a backpropagation (BP) neural network algorithm. The research results showed that under the same conditions, when parking points were arranged according to the conventional method, the sharing cycle utilization rates of stations and residential areas at 7–9, 11–13 and 17–19 were 92.6% and 90.4%, 88.9% and 87.6%, 89.3% and 85.5% respectively; when parking points were arranged according to big data and BP neural network algorithm, the utilization rates of shared bicycles at 7–9, 11–13 and 17–19 in stations and residential areas were 95.1% and 94.3%, 91.9% and 90.2%, 93.7% and 94.1% respectively. The utilization rate of the latter was significantly higher than that of the former, indicating that big data and neural network methods can improve the utilization rate of shared bicycles.<br><br>Keywords: Bike sharing; neural network method; big data; rail transit station<br><br>Cite As<br><br>Y. Shao, T. Zhang, M. Hong, F. Zhou, J. Zhou, "A Study on the Utilization of Shared Bicycles in Comprehensive Passenger Transport Hub Area Based on Neural Network Evaluation", <em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 579-592, 2026.</p> <p><br><br><br><br></p> Yunhong Shao, Tao Zhang, Min Hong, Feiqiang Zhou, Jing Zhou Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2048 Tue, 01 Sep 2026 00:00:00 -0700 Application of Natural Language Processing Technology in Multilingual Sentiment Analysis and Sentiment Recognition https://website-eis.crlpublishing.com/index.php/eis/article/view/2049 <p>The recent increase in social media data in multiple languages has led to fragmented emotional utterances that contain dialects, slang, and emojis. Traditional NLP (Natural Language Processing) models are ineffective in detecting cross-lingual implicit emotions; thus, there is a need to develop a robust multimodal multilingual emotion recognition framework. This paper introduces a multilingual emotion understanding framework. First, the study utilizes XLM-R for cross-lingual semantic alignment and employs parallel corpus comparison learning to ensure the aligned encoding of emotional utterances in different languages within the same space. Second, a multimodal emotion symbol fusion mechanism is employed, utilizing CLIP and Emoji2Vec, to better represent the semantic meaning of emojis and images. Third, fastText and a dynamic dictionary mechanism are applied to adapt to dialects and slang, which facilitates the recognition of emotional utterances and non-standard texts. Fourth, adversarial training as part of transfer learning is used to increase cross-lingual robustness and emotion consistency. Lastly, BERT-CRF combined with an attention mechanism is utilized for joint recognition of polarities and fine-grained emotions, leading to better detection of sarcasm and implicit emotional understanding. Experiments show that the model achieves an accuracy of 89.2% in English emotion recognition across five languages, and the accuracy of dialect recognition in high-resource dialects is improved to 90.2% through a dynamic mechanism. With an overall accuracy of 82% and attention coverage of 69.3%, the framework’s effectiveness in cross-language, multimodal, and non-standard texts was validated.<br /><br />Keywords: Multilingual Emotion Recognition; Cross-lingual Semantic Alignment; Multimodal Emotion Fusion; Adversarial Transfer Training; Fine grained Emotion Classification<br /><br />Cite As<br /><br />Y. Yuan, "Application of Natural Language Processing Technology in Multilingual Sentiment Analysis and Sentiment Recognition", <em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 593-606, 2026.</p> <p><br /><br /><br /></p> Yang Yuan Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2049 Tue, 01 Sep 2026 00:00:00 -0700 Collaborative Fault Prediction and Intelligent Localization of Distributed Operation and Maintenance Platform Based on Spatiotemporal Graph Neural Network https://website-eis.crlpublishing.com/index.php/eis/article/view/2050 <p>Traditional approaches to Distributed Systems Operations and Maintenance cannot adequately predict faults or locate root causes. To solve these problems, this paper presents a Spatio-Temporal Graph Neural Network (STGNN) as an intelligent framework. The STGNN framework leverages dynamic graphs for both the topological and the state evolution models of the distributed system. The prediction component utilizes deep learning algorithms. The localization component applies cross-node attention and graph reasoning to automatically reconstruct the chain of fault propagation and identify the node that caused the fault. Experimental results indicate that the proposed approach achieves a fault prediction accuracy of 90.2%, and reduces the average root cause localization time for write latency in databases by 35.6%. This paper provides a more effective and reliable means of enabling intelligent operations and maintenance in distributed systems.<br /><br />Keywords: distributed operations management; spatio temporal graph neural network; dynamic graph structure; cross-node attention; fault propagation chain<br /><br />Cite As<br /><br />J. Ma, T. Fu, M. Lei, Z. Zhou, L. Liang, "Collaborative Fault Prediction and Intelligent Localization of Distributed Operation and<br />Maintenance Platform Based on Spatiotemporal Graph Neural Network", <em>Engineering Intelligent Systems,</em> vol. 34 no. 5, <br />pp. 607-620, 2026.</p> <p><br /><br /><br /><br /></p> Jianwei Ma, Tongfu Fu, Ming Lei, Zhongqiang Zhou, Ling Liang Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2050 Tue, 01 Sep 2026 00:00:00 -0700 Content Analysis of Short Video Dissemination From the Perspective of College Students’ Preferences https://website-eis.crlpublishing.com/index.php/eis/article/view/2051 <p>This study was conducted to determine the extent to which students use short videos. The study subjects were college students from the Beijing, Jiangsu, Shandong, and Henan provinces in China. A questionnaire survey was used to assess students’ reliance on short videos. A crawler program was used to collect students’ browsing history of short videos to investigate the different types preferred by students. The K-means algorithm was used to classify students based on their level of addiction to short videos, and the short video dissemination preferences of different classifications were compared. It was found that the surveyed college students showed a preference for short videos of situational skits, creative clips, and news highlights. There were 2,017 individuals in the normal group and 879 individuals in the addiction group. Survey results showed that the addiction group demonstrated significantly higher average viewing duration, number of forwards, number of likes, and number of collections for short videos compared to the other group.<br><br>Keywords: short video, college student, preference tendency, dissemination<br><br>Cite As<br><br>J. Zhang, Z. Li, "Content Analysis of Short Video Dissemination From the Perspective of College Students’ Preferences",<br><em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 621-625, 2026<br><br><br><br><br></p> <p>&nbsp;</p> Jianbao Zhang, Zhengang Li Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2051 Tue, 01 Sep 2026 00:00:00 -0700 Metaverse-oriented Multi-task Learning Algorithms in Animation Scene Design https://website-eis.crlpublishing.com/index.php/eis/article/view/2052 <p>In response to the problems of poor efficiency in animation scene design and the inadequate immersive experience of users , this study focuses on the metaverse and takes the animation Spirited Away as the object, applying multi-task learning algorithms to examine animation scene design. Firstly, a multi-task learning framework based on generative adversarial network (GAN) is constructed. The deep convolution generative adversarial networks method (DCGAN) is used as the basic architecture of the generator, and the convolutional neural network (CNN) is used as the basic architecture of the discriminator to model and generate the environment of the animation scene. Then, physically-based rendering (PBR) technology is combined to simulate the propagation and reflection of light in the scene for lighting modeling. Percentage closer filtering (PCF) technology is used to process soft shadows so as to enhance the realism of the animation scenes. The Monte Carlo path tracing algorithm is used to comprehensively calculate the global lighting effects of various areas in the scene. Finally, the Tone Mapping algorithm is used for color space conversion and mapping processing, and efficient rendering and realistic effects are demonstrated through deferred rendering and dynamic lighting processing techniques, resulting in the final dynamic scene generation rendering. The experiment is divided into two tasks: dynamic environment modeling and light and shadow effect design and optimization. MT-DCGAN (multi-task deep convolutional generative adversarial network) achieves a design efficiency of 95.73%, with an improvement of 13.49% compared to that of DCGAN, and a user’s attention concentration of 93.36%. The combination of multi-task learning algorithms in animation scene design greatly improves design efficiency and user experience, and enhances user satisfaction with animation scenes, thereby contributing an important innovative approach to the animation industry.<br /><br />Keywords: Metaverse direction; animation scene design; multi-task learning algorithm; Spirited Away; immersive experience<br /><br />Cite As<br /><br />Z. Song, X. Guo, Y. Gu, "Metaverse-oriented Multi-task Learning Algorithms in Animation Scene Design", <em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 627-639, 2026<br /><br /></p> Zhuoyi Song, Xinyu Guo, Yu Gu Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2052 Tue, 01 Sep 2026 00:00:00 -0700 Precise Recommendation of Sichuan Cultural and Tourism Information in Short-Video Apps Based on User Profiles https://website-eis.crlpublishing.com/index.php/eis/article/view/2053 <p>This paper briefly introduces the construction of user profiles and a user profile-based short video recommendation algorithm. In the algorithm, the user profiles are first classified using the clustering algorithm improved by particle swarm optimization (PSO), and then short videos are recommended to users based on the profiles. Simulation experiments were then conducted on the Douyin short video platform. The results showed that the PSO improved K-means clustering algorithm effectively classified user profiles, with the best classification effect when the K value was 4. The proposed recommendation algorithm could provide users with more accurate short videos on Sichuan culture and tourism. <br /><br />Keywords: User profile, short video, recommendation algorithm, clustering algorithm</p> <p><br />Cite As<br /><br />Y. Hu, "Precise Recommendation of Sichuan Cultural and Tourism Information in Short-Video Apps Based on User Profiles", <em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 641-646, 2026.<br /><br /></p> <p><br /><br /><br /></p> Yangjian Hu Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2053 Tue, 01 Sep 2026 00:00:00 -0700 Recognition of Movement Features of Different Ethnic Dances: Research on Motion Image Information https://website-eis.crlpublishing.com/index.php/eis/article/view/2054 <p>The inheritance and teaching of ethnic dances can be innovated by introducing digital technology. This paper studied the recognition of movement features of different ethnic dances. It optimized the spatiotemporal graph convolutional network (ST-GCN) based on skeleton information in motion images, developed a multi-stream ST-GCN (MS-STGCN) model, and conducted experiments using NTU-RGB+D60/120 datasets and a self-constructed ethnic dance dataset. The results showed that for the NTU-RGB+D 60 dataset, the accuracy of the MS-STGCN model was 92.6% (X-Sub) and 96.8% (X-View); for the NTU-RGB+D 120 dataset, its accuracy was 89.7% (X-Sub) and 89.9% (X-Set), outperforming the other algorithms; it achieved an accuracy of 90.6% for the ethnic dance dataset. The findings demonstrate the reliability of the MS-STGCN model in recognizing the features of ethnic dance movements, which can be further applied in practice.<br /><br />Keywords: spatiotemporal graph convolutional network; ethnic dance; motion image; recognition<br /><br />Cite As<br /><br />F. Yang, "Recognition of Movement Features of Different Ethnic Dances: Research on Motion Image Information",<br /><em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 647-652, 2026.</p> <p><br /><br /><br /><br /><br /></p> Fan Yang Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2054 Tue, 01 Sep 2026 00:00:00 -0700 Research on Closed Loop Management Information System of Hybrid Teaching Driven by Generative Artificial Intelligence https://website-eis.crlpublishing.com/index.php/eis/article/view/2055 <p>&nbsp;</p> <p>To address significant problems such as feedback lag, insufficient personalization and static teaching resources in hybrid teaching, this paper proposes a closed-loop adaptive generative teaching engine—CLAGTE—driven by generative artificial intelligence. The engine consists of multi-modal student state awareness network MHAFNet, time-series attention knowledge tracking model TAKT, teaching strategy decision module based on near-end strategy optimization and retrieval enhancement generative content creation module, which form a full-link closed loop of “perception-diagnosis decision-generation”. Experimental results on three public datasets of ASSISTments, Junyi Academy and EdNet show that the AUC of CLAGTE in<br>performance comparison is as high as 0.879, which is about 2.6 percentage points higher than the optimal baseline DIMKT. Its AUC decreases by only 11.9% under 30% label noise, and its robustness is significantly better than that of the comparative model. At the same time, the comprehensive mean of practical subjective score reaches 4.33, and the interpretable Spearman correlation coefficient is 0.68. The accuracy rate of learning gain simulation increases by 7.8% after a single intervention, and the cumulative accuracy rate increases by 14.6%. In addition, its cross-domain zero sample migration AUC remains above 0.75. Furthermore, the ablation experiment verifies the effectiveness of each module. In summary, CLAGTE achieves the intelligent closed loop of hybrid teaching management through the collaboration of multi-modal fusion, dynamic knowledge tracking, reinforcement learning adaptive strategy and generative content supply, and provides a feasible technical solution for personalized education.<br><br>Keywords: Generative artificial intelligence; hybrid teaching; closed-loop management information system; knowledge tracking; multimodal fusion<br><br>Cite As<br><br>J. Zhang, X. Wang, "Research on Closed Loop Management Information System of Hybrid Teaching Driven by Generative Artificial Intelligence", <em>Engineering Intelligent Systems,</em> vol. 35 no. 5, pp. 653-664, 2026.<br><br></p> <p><br><br><br><br><br><br></p> Jian Zhang, Xiu Wang Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2055 Tue, 01 Sep 2026 00:00:00 -0700 Statistical Analysis and Advanced Predictive Techniques to Optimize Detection of Network Traffic Anomaly and Improve Network Security and Threat Identification https://website-eis.crlpublishing.com/index.php/eis/article/view/2056 <p>Traffic patterns in today’s network environment vary considerably, and traditional anomaly detection methods were not designed to address the various issues relating to the high dimensionality of structural representations of network topologies, temporal dependencies, and real-time response, leading to decreased accuracy and consistency in identifying anomalies. The research presented in this paper proposes a novel method of detecting anomalies within network traffic utilizing statistical analysis optimized by means of advanced predictive techniques. Using the structured features of cleaned-up network traffic datasets, the study creates a statistical feature analysis approach in order to develop a baseline of normal network traffic behaviour; this provides insight into the statistical nature of your organisation’s network traffic patterns. In addition, the paper utilizes an LSTM and a multi-level predictive model to develop a methodology to predict the future behaviour of the network traffic. Finally, by leveraging the residual-based anomaly scoring techniques, experiments demonstrate the ability to support accurate and timely identification of network traffic anomalies. The experiments were conducted on many different types of network traffic datasets and resulted in an anomaly detection accuracy of 96.8%, an F1 score of 95.9%, a false positive rate of 0.02, a false negative rate of 0.11, and an average residual prediction convergence rate of 0.0125 during training. The median time required for the detection of anomalies end-to-end was 0.20 seconds. The results presented in this paper verify that the proposed methodology has the potential for very precise and low-latency detection of network traffic anomalies and will provide technical support for the identification of network security and network threat risks.<br /><br />Keywords: Network traffic anomaly detection; advanced prediction techniques; statistical analysis; network security; threat identification<br /><br />Cite As<br /><br />Z. Zhang, X. Liu, "Statistical Analysis and Advanced Predictive Techniques to Optimize Detection of Network Traffic Anomaly and Improve Network Security and Threat Identification", <em>Engineering Intelligent Systems,</em> vol. 35 no. 5, pp. 665-681, 2026.</p> <p><br /><br /><br /></p> <p> </p> Zhe Zhang, Xiangjiao Liu Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2056 Tue, 01 Sep 2026 00:00:00 -0700 SymLoop: A Symbolic Constraint–Integrated Framework for Consistency Preservation and Reasoning Control in English Text Generation https://website-eis.crlpublishing.com/index.php/eis/article/view/2057 <p><br />Large language models (LLMs) have demonstrated remarkable performance in English text generation. However, their prevailing next-token prediction paradigm struggles to internalize deep causal reasoning rules, often resulting in outputs that are formally well-structured yet logically fragmented. Existing approaches suffer from notable limitations: declarative prompting methods lack deep integration with logical rules; external solver-based approaches rely on sequential pipelines that impede real-time correction; and constrained decoding methods disrupt linguistic fluency due to discrete interventions. To address these challenges, we propose SymLoop, a symbolic constraint-integrated framework for consistency preservation and reasoning control in English text generation. SymLoop establishes a three-stage neuro-symbolic closed loop of extraction- injection- verification.<br />Specifically, we introduce a Context-Adaptive Symbolic Constraint Extraction (CASCE) algorithm that dynamically captures implicit logical entailments; a Bayesian Symbolic Constraint Injection (BSCI) algorithm that softens discrete rules into probabilistic priors within the latent space; and a Hierarchical Symbolic Verification Feedback Loop (HSVFL) that enables end-to-end logical correction through reinforcement learning. Experimental results show that SymLoop achieves a logical consistency accuracy of 0.94 and a BERTScore of 0.89 on ProofWriter, and outperforms the strongest baseline by 12% on EntailmentBank. The reasoning latency is 48 ms/token, yielding a 2.6× speedup over classical constrained decoding methods. This work introduces a new paradigm for integrating probabilistic modeling with logical reasoning in neural language models, providing technical support for<br />applications such as academic writing and the drafting of legal documents.<br /><br />Keywords: large language models; neuro-symbolic integration; logical consistency; English text generation; variational inference; SymLoop<br /><br />Cite As<br /><br />X. Liu, "SymLoop: A Symbolic Constraint–Integrated Framework for Consistency Preservation and Reasoning Control in English Text Generation", <em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 683-696, 2026.<br /><br /></p> <p><br /><br /><br /><br /><br /></p> Xiaoyi Liu Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2057 Tue, 01 Sep 2026 00:00:00 -0700 User Profile: Research on Employment Recommendation and Guidance for College Graduates After Graduation https://website-eis.crlpublishing.com/index.php/eis/article/view/2058 <p>As the number of college graduates continues to increase, the employment pressure has also increased. Based on user profiling, this paper designed a post-graduation employment recommendation and guidance method for college students. The method extracted features such as work experience and salary from college students’ resumes and job information, used the K-means++ algorithm to cluster the resumes and job information respectively to obtain user and job profiles, and matched the resume text with the job text to obtain the recommendation list. Student data from a college in Sichuan Province, China, were selected for experiments. It was found that the K-means++ algorithm outperformed the traditional algorithm in clustering, with<br>a Calinski-Harabasz index of 503 and a silhouette coefficient of 0.812. In terms of employment recommendation, when the number of recommended positions was 20, the recommendation effect of this method was the best, with an F1 value of 0.655, outperforming the traditional collaborative filtering algorithm. The results demonstrate the reliability of the proposed method in employment recommendation and guidance, and it can be applied to the employment of post-graduates.<br><br>Keywords: user profiling, college student, employment recommendation, k-means++<br><br>Cite As<br><br>J. Luo, "User Profile: Research on Employment Recommendation and Guidance for College Graduates After Graduation",<br><em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 697-702, 2026.</p> <p><br><br><br><br><br></p> Jing Luo Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2058 Tue, 01 Sep 2026 00:00:00 -0700 Using the Decision Tree Algorithmto Explore the Factors Influencing Social Anxiety Among University Students https://website-eis.crlpublishing.com/index.php/eis/article/view/2059 <p>This study is conducted to analyze the factors influencing social anxiety among university students using the decision tree (DT) algorithm to provide some bases for its intervention. Based on convenient sampling, questionnaires were distributed to students of a university, and 1,832 valid responses were obtained. The questionnaire included items eliciting demographic information, and the interaction anxiousness scale (IAS). Logistic regression and chi-squared automatic interaction detector (CHAID) were used to analyze the questionnaire data. Data analysis revealed that 28.28% of the students experienced social anxiety. Both the Logistic and DT models showed that family structure completeness, family financial status, leadership qualities, and whether one is an only child were factors influencing the social anxiety of students. The comparison found that the area of the curve under the receiver operator characteristic of the DT model was larger than that of the Logistic model (Z = 9.264, p &lt; 0.01), indicating that the<br>decision tree algorithm performs well in determining the factors influencing social anxiety. More attention should be paid to students with incomplete family structures and those whose families are not financially secure, so as to help them establish confident and acceptable social behaviors.<br><br>Keywords: Decision tree, social anxiety, influencing factor, family<br><br>Cite As<br><br>N. Yao, "Using the Decision Tree Algorithmto Explore the Factors Influencing Social Anxiety Among University Students",<br><em>Engineering Intelligent Systems,</em> vol. 34 no. 5, pp. 703-708, 2026.</p> Na Yao Copyright (c) 2026 International Journal of Engineering Intelligent Systems https://website-eis.crlpublishing.com/index.php/eis/article/view/2059 Tue, 01 Sep 2026 00:00:00 -0700