International Journal of Web Engineering and Technology (18 papers in press)

Regular Issues

• Service Recommendation Method based on Text View and Interaction View  
  by Shuaijia Lin, Ting Yu, Yaqi Wang, Jie Xu, Fangying Cheng, Tian Liang 
  Abstract: With the increasing prosperity of web service-sharing platforms, more and more software developers are reusing web services when developing applications. Existing web service recommendation systems often face two challenges. Firstly, developers discover services by inputting requirements, but the user's input is arbitrary and it cannot fully reflect the user's intention. Secondly, the application-service interaction records are too sparse, making it particularly difficult to find services that meet the requirements. To address the above challenges, in this paper, we propose a service recommendation method based on text and interaction views (SRTI). Firstly, SRTI employs graph neural network to deeply mine the features of applications and services. Secondly, SRT uses transformer and fully connected neural networks to deeply mine the matching degree between candidate services and requirements. Finally, we integrate the above two to obtain the final service list. Extensive experiments on real-world datasets have shown that SRTI outperforms several state-of-the-art methods.
  Keywords: service recommendation; text view; interaction view; application; recommendation algorithm.
  DOI: 10.1504/IJWET.2024.10064249
   
  
• Performance and Reliability Analysis of Web Service Composition Using Probabilistic Model Checking  
  by Khadidja Salah Mansour, Youcef Hammal 
  Abstract: Design and implementation of composite software systems by integrating loosely coupled individual Web services are still an important issue within the SOA topic even though they still face significant complexities related to compatibility challenges which arise from their independent development processes. Our key aim is hence to provide an in-depth formal analysis of compatibility issues encountered in composite service and propose a comprehensive methodology to effectively address these challenges. We propose a formalisation and verification approach to tackle the crucial problem of services integration that always hinders reliable composition of distributed services. This paper focuses on ensuring functional and non-functional correctness of composite services described using WS-BPEL. These descriptions are first translated into communicating automata and then checked against intended specifications defined by WS-CDL. In order to enable probabilistic model checking of the composite service using the PRISM tool, the external choreography properties are formalised into formulae of PCTL.
  Keywords: Service Web composition; PTA; Model checking; CDL; BPEL; PCTL;SOA.
  DOI: 10.1504/IJWET.2026.10071686
   
  
• Visualisation of Global Research Trends and Future Research Directions in Search Engine Marketing Research: a Bibliometric Analysis (2004-2024)  
  by Prachi Kapil, Prince Vohra, Vikas Deep, Seema Singh 
  Abstract: Over the past two decades, search engine marketing (SEM) has evolved into a critical area of digital marketing, yet a comprehensive understanding of its research landscape remains fragmented. This study addresses this gap by conducting a bibliometric analysis of 246 academic articles published between 2004 and 2024, sourced from the Scopus database. Utilising bibliometric tools, we identify key research trends, influential publications, leading authors, and prominent journals shaping SEM scholarship. Our analysis uncovers thematic clusters within SEM research, revealing its progression from foundational SEO and keyword advertising to AI-driven personalised marketing strategies. Furthermore, we map global contributions, highlighting top institutions and nations driving research advancements. By systematically analysing citation patterns and keyword co-occurrences, this study not only synthesises existing knowledge but also identifies future research directions, ensuring SEM remains adaptive to the evolving digital ecosystem.
  Keywords: Search engine marketing; SEM; bibliometric analysis.
  DOI: 10.1504/IJWET.2026.10072471
   
  
• High Quality Video Creation that Integrates Audio and Visual Feature Information  
  by Hua Fan, Dongdong Lin 
  Abstract: Video generation involves creating virtual video images through computer programs with diverse applications. This study focuses on cross-modal video generation from audio to visual information. It extracts Mel cepstral coefficient features from audio signals and establishes a convolutional neural network with self-attention. A high-quality video creation model is developed based on neural radiation field representation. Results indicate an Inception score around 1.6 to 1.7, showing good convergence of the loss function at 40000 training frequency. The proposed algorithm enhances peak signal-to-noise ratio by 24.7% compared to standard methods, with training speed improved by up to 1-2 times. The model exhibits resolution flexibility, robustness across scenarios, and improved video viewing experience compared to existing methods, promising advancements in video creation technologies.
  Keywords: Audio features; Visual features; Video creation; Mel cepstral coefficient characteristics; Convolutional neural network.
  DOI: 10.1504/IJWET.2026.10072544
   
  
• Tracking and Monitoring Framework using Context-based Approach for Real Time Surveillance: a Case Study  
  by Nivedita Ray De Sarkar, Anirban Kundu, Mou De 
  Abstract: We propose a monitoring system based on real-time surveillance for green society with internet of things (IoT) for understanding location of user/substance in enclosed environment. Tracking individuals and locations using maps, provided by different organisations, have been seen several times. We aim to design tracking within specified enclosed area. All users entering the proposed system are provided with a device to generate live data of their movements within the area. Generated live data is sent over the web for further tracking based on pre-defined context. New theory based on total displacement and its relation to initial position, final position and number of displacements has been proposed in this paper. The system aims towards contributing to green society providing reusable, low emission devices for tracking. Our system is distributed with server side being remotely located for overall analysis. We are going to formulate the tracking system by experimenting on different contexts (circle mesh, triangle mesh, hexagon mesh, octagon mesh). Assurance factor and percentage of change in prediction are used to predict accurate results. Comparison with existing surveillance systems show improved results.
  Keywords: Tracking; IOT; Surveillance; context awareness; security; system monitoring; Green Computing; Green Society; Closed-circuit Television (CCTV).
  DOI: 10.1504/IJWET.2026.10072673
   
  
• Construction of Online English Teaching Quality Evaluation Model based on PCA-SSA-SVM Algorithm  
  by Jingjia Guo, Lei Zhang, Naiyuan Bao 
  Abstract: This study proposes a hybrid algorithm for objective online English teaching quality evaluation. principal component analysis (PCA) reduces feature dimensions and extracts key components, sparrow search algorithm (SSA) optimises support vector machine (SVM) parameters, and SVM performs classification/prediction. Experiments on the PISA dataset demonstrate the models superiority: optimal convergence in five iterations, 1% average classification error, 98.7% fit degree, and 45.2s training time, outperforming other methods. The algorithms stability, accuracy, and efficiency enable rapid model updates, aiding timely teaching strategy adjustments. Results confirm its enhanced problem-solving, pattern recognition, and generalisability, offering a robust solution for standardised, data-driven quality assessment in online education.
  Keywords: SVM; SSA; PCA; Online English language teaching; Quality assessment.
  DOI: 10.1504/IJWET.2026.10072718
   
  
• Wireless Network Vertical Handover Algorithm Based on FAHP-DDQN  
  by Huan Wang 
  Abstract: To improve switching efficiency in heterogeneous wireless networks, this study proposes a novel vertical switching algorithm. Combining fuzzy tomographic analysis with traditional vertical switching techniques, a wireless network vertical switching algorithm based on fuzzy analytic hierarchy process is designed. Further optimisation is achieved by integrating a dual deep Q-network. The optimised algorithm requires only 42 iterations to achieve the optimal strategy, achieving a high resource utilisation rate of 98.2% in personal area networks with a response time of 0.3 seconds. This optimised vertical switching algorithm demonstrates good benchmark performance and practical application results, providing technical support for efficient data switching in wireless networks.
  Keywords: AHP; Deep Q-network; Heterogeneous network; Wireless communication; Vertical switching.
  DOI: 10.1504/IJWET.2026.10073138
   
  
• A method for generating animated character expressions based on improved MTCNN  
  by Zhongkai Zhan 
  Abstract: With the growth of the domestic animation industry, character expression generation has diversified. To improve naturalness and coordination, this study proposes an improved multi-task convolutional neural network. The method uses a hierarchical facial data structure with cascaded network features optimised by non-maximum suppression, bounding-box regression for feature range control, and facial landmark recognition to stabilise spatial information. Action units enhance the correlation between facial expressions and partition actions, while a graph convolutional network and weighted binary cross-entropy loss optimise results. Experiments show fast convergence, stable performance across datasets, and superiority over traditional models such as the active appearance model. Structural similarity index values reach 0.863 for large amplitude and 0.931 for small amplitude expressions. The algorithm generates coherent, natural, and authentic composite expressions, offering an effective model for animated character expression synthesis.
  Keywords: Character expressions; Multi-task convolutional neural network; Non-maximum suppression; Action unit; Structural similarity index.
  DOI: 10.1504/IJWET.2026.10073460
   
  
• E-commerce Customer Churn Analysis and Multi-Algorithm Fusion Prediction Study  
  by Zhuqing Ji 
  Abstract: With the rapid development of Internet and e-commerce, customer churn has become an important challenge for the platform. Accurate prediction of churn helps enterprises to intervene in time and improve customer retention and profitability. This paper proposes a multi algorithm model that combines unsupervised and supervised learning. Unsupervised learning uses K-means++, K-centres and two-step clustering to group customers to improve the quality of features. Supervised learning integrates logistic regression, support vector machine, back propagation neural network and C4.5 decision tree for classification. The experimental results show that the specificity is 0.98, the sensitivity is 0.73, the accuracy is 0.93, the accuracy is 0.77, the G mean is 0.86, and the F value is 0.76. The model can effectively predict the loss of customers and has low misjudgement rate. Enterprises can formulate personalised customer retention strategies to reduce customer churn and improve long-term profitability
  Keywords: K-means; EasyEnsemble-Smote; Multi-algorithm fusion; E-commerce.
  DOI: 10.1504/IJWET.2026.10073834
   
  
• Fingerprint Evidence Recognition System Based on Adaptive Network Fuzzy Inference System  
  by Hui Sun 
  Abstract: This paper introduces an adaptive network fuzzy inference system. Image preprocessing is performed through Gaussian filtering, histogram equalization and morphological operations. Fingerprint features are extracted using directional gradient histogram and Gabor filter. ANFIS model is constructed, fuzzy rules are designed, and membership parameters are optimized through a backpropagation algorithm to calculate the matching degree of feature points. It is combined with a support vector machine for identity recognition or evidence classification. The fingerprint evidence recognition method based on the adaptive network fuzzy inference system significantly improves the recognition accuracy. It enhances the system's adaptability and robustness by combining image preprocessing and feature extraction technology, providing new ideas and technical support for the practical application of fingerprint evidence recognition technology.
  Keywords: Biometric Technology; Fingerprint Evidence Recognition; Adaptive Network Fuzzy Inference System; Feature Extraction; Image Preprocessing.
  DOI: 10.1504/IJWET.2026.10074175
   
  
• Financial Risk Control Model Integrating BERT and Long Short-Term Memory Networks  
  by Xueli Deng, Yi Li 
  Abstract: With the continuous evolution and transformation of financial markets, financial risks are becoming increasingly diversified and complex. Traditional risk control methods are often limited by the ability of feature extraction and temporal modelling, making it difficult to accurately identify and predict financial risks. Therefore, the study integrates the bidirectional encoder transformation representation model with the long short-term memory network, utilises the bidirectional encoder transformation representation model for feature extraction in financial risk control, and constructs a hybrid financial risk control model that can learn the mapping relationship between features and risks. Results showed that the accuracy, rejection rate, and false acceptance rate of the mixed model were 91.93%, 5.76%, and 9.36%, respectively, proving that its financial risk classification accuracy was relatively high. The Kappa coefficients of the hybrid model were 16.4%, 32.84%, and 54.37% higher than those of other models, indicating that it had higher classification ability, discrimination ability, and consistency. The designed model has shown high performance in financial risk control, significant for financial markets stable operation and systemic risks prevention.
  Keywords: Finance; BERT; Long short-term memory network; Hybrid model; Control; Risk management.
  DOI: 10.1504/IJWET.2026.10074178
   
  
• Enhancing AI Teaching Methods in CoI Scenarios through Emotion-Aware Interaction  
  by Si Chen, Lina Wang 
  Abstract: The rapid expansion of artificial intelligence in education is prominent. This study focuses on anthropomorphizing AI to refine teaching approaches. By leveraging deep interactive semantic alignment and self-explanatory graph attention networks, AI comprehension of subtle human language nuances is enriched. The research fosters learner-AI cohesion, constructing an inquiry community to evaluate teaching data. Model evaluation yielded promising results: Accuracy 75.12%, Precision 56.05%, Recall 50.67%, and F1 54.17%. In comparison, the RERT model scored Accuracy 72.49%, Precision 55.95%, Recall 51.07%, and F1 53.45%. Aligning with Community of Inquiry theory, students' affinity for AI imagery correlates positively with various presences. Such imagery influences learner-AI dynamics significantly, emphasizing the pivotal role of anthropomorphic AI in education.
  Keywords: Community of inquiry; Artificial intelligence; Anthropomorphism; Graph attention network; Teaching.
  DOI: 10.1504/IJWET.2026.10074179
   
  
• ES-YOLO: Research on Security Protective Equipment Detection Based on Improved YOLOv8.  
  by Shijie Guan, Zihan Li 
  Abstract: In the construction industry, safety protective equipment is crucial for protecting the lives and property of workers. However, due to the complex background of construction sites and the diverse scales of targets, the targets are prone to be neglected or falsely detected. To alleviate the above problems, we propose an improved YOLOv8 algorithm called ES-YOLO. Firstly, to better learn contextual information in complex scenes, Efficient Multi-Scale Attention mechanism is introduced after every C2f module in the neck of YOLOv8. Secondly, we introduce a new Small Target Detect Layer to improve the performance of algorithm for multi-scale target detection. Finally, the experiment shows that P, R, and mAP are significantly improved by 3.1%, 1.7%, and 2.5%, which proves the effectiveness of ES-YOLO in alleviating the problem of target misdetection and omission.
  Keywords: EMA; efficient multiscale attention mechanism; YOLOv8; computer vision; safety protective equipment detection.
  DOI: 10.1504/IJWET.2026.10074367
   
  
• Intelligent Logistics Service Solution for IoT Distribution and Dispatch activities  
  by Qijun Ren 
  Abstract: On the other hand, it does not produce the effect that was anticipated and operates with a low level of optimizing processing efficiency. It is recommended that an Internet of Things-based logistics collaborative distribution route optimisation solution be used in order to resolve the problem. As part of the collaborative logistics distribution and dispatch system, Internet of Things technology has been included into the logistics distribution process. This integration allows for the use of dynamic factors as limits on route optimization. Following the establishment of the collaborative distribution network, the logistics route optimisation model is constructed and then solved with the assistance of the scanning-genetic algorithm in order to accomplish the ultimate outcomes. According to the results of the studies, the new method reduces the number of vehicle miles travelled by an average of 14.5% by the reduction of distribution costs and the enhancement of efficiency.
  Keywords: IoT technology; logistics distribution and dispatch; collaborative distribution; path optimisation.
  DOI: 10.1504/IJWET.2026.10074712
   
  
• Multi-Data Dimension Bisecting Clustering for user Profile Analysis based on Library User Information  
  by Changsheng Wang, Chuanyu Zhang 
  Abstract: With the rapid development of internet technology, library users demand for books is more obvious. However, how to improve the quality of library services to users and provide more accurate information and personalised services has become an important issue in current library management and user access. Therefore, a user multi-data dimension model based on bisecting clustering algorithm is proposed in the study, which improves the accuracy of recommending user profiles by improving the bisecting clustering algorithm. The research results indicate that the new model consumes 75 seconds less energy than K-means algorithm, and the distance values of parameters are also lower than other models. The research algorithm has the highest accuracy and F1 value, 0.26 and 0.13 higher than the other algorithms. From this, it can be seen that the performance and recommendation accuracy of research method have good results, which has good guiding value for accurate analysis of library user profiles and recommendation research in the future.
  Keywords: Library; User profile; Multi-data dimension; Bisecting clustering; Recommendation.
  DOI: 10.1504/IJWET.2026.10074762
   
  
• Job Recommendation Systems and the Benefits from the use of Gallup Tests  
  by Shakhmar Sarsenbay, Iraklis Varlamis, Cemil Turan, Zhadyra Zhalgassova 
  Abstract: Job recommendation systems play a vital role in modern recruitment processes by leveraging data analytics and machine learning algorithms to provide personalised and relevant job recommendations to both job seekers and employers. This work examines how job recommendation systems work, focusing on their key components, algorithmic techniques, data sources and evaluation metrics. Furthermore, it explores the potential of these systems on taking advantage of employee profiling services, improving the efficiency of skill matching for individuals, and enhancing the recruitment processes for organisations. Additionally, ethical considerations and real-world examples of job recommendation systems are discussed, shedding light on their effectiveness, limitations and future developments. Finally, the article proposes a novel approach for using employee profile information and more specifically their profile features as described by Gallup tests in conjunction with their CV information and the information of a job posting they are interested in, in order to recommend them which skills to improve and help companies to improve job postings in order to better state the required soft skills.
  Keywords: Job Recommendation Systems; Employee Profiling; Skill Matching; Recruitment Processes; Gallup Tests; Soft Skills.
  DOI: 10.1504/IJWET.2026.10075161
   
  
• Teaching Classroom Learning Behaviour Analysis Based on Support Vector Machine and CLM  
  by Chunlin Xiang, Lingtong Li, Yan Jiang 
  Abstract: Learning behaviour reflects student engagement and knowledge absorption. However, current automatic recognition methods lack accuracy and efficiency. To enhance teaching quality and provide precise data, a model based on support vector machine and Constrained local model is proposed. Visual sensors and cameras capture student motion and expressions. Facial expressions and behavioural features are extracted using local binary patterns and bone data. Support vector machines and facial tracking classify behaviours. It provides teachers with a basis for judging the quality of teaching based on the identification results. The model achieves over 95% accuracy and runs in 0.2s, ensuring real-time analysis. From this, the designed model can more accurately capture subtle differences and potential patterns in learning behaviour, provides teachers with insights to optimise teaching strategies, enhancing student learning effectiveness.
  Keywords: SVM; Learning behavior analysis; Kalman filtering; Teaching quality; Emotional recognition.
  DOI: 10.1504/IJWET.2026.10075390
   
  
• Exploration of New Media Film Special Effects Based on Computer Vision and Virtual Reality Technology  
  by Yawen Tang, Jianhong Ren 
  Abstract: Traditional movie special effects have a single colour palette and cannot meet the needs of the audience. This paper analyses new media movie special effects based on CV and VR technology, selecting three indicators: realism, user experience, and stability of movie special effects, and selecting user experience for analysis. The experimental results show that the average scores of immersion, virtuality, and interactivity in traditional user experience are 60.5, 64, and 42.5, respectively, while the average scores of VR user experience in these three aspects are 75, 79, and 85.5, respectively. The average user experience score of VR technology is higher than that of traditional methods. Immersion, virtuality, and interactivity are the standards for measuring users' viewing experience of new media movie special effects. The higher the score, the stronger the viewing effect. The combination of VR technology and new media movie special effects can better cater to users' viewing experience.
  Keywords: New Media Film Special Effects; Virtual Reality Technology; Computer Vision; Image Preprocessing.
  DOI: 10.1504/IJWET.2026.10075725