Systems and Soft Computing最新文献

{"title":"Research on the construction and application of intelligent tutoring system for english teaching based on generative pre-training model","authors":"Weiguo Feng ,&nbsp;Xinhan Lai ,&nbsp;Xin Zhang ,&nbsp;Xiaojing Fan ,&nbsp;Yinxia Du","doi":"10.1016/j.sasc.2025.200232","DOIUrl":"10.1016/j.sasc.2025.200232","url":null,"abstract":"<div><div>In the era of globalization and information technology, English is becoming more and more important as the main language of international communication, but the traditional English teaching model is difficult to meet the diverse needs of learners due to the uneven distribution of resources and the lack of personalized tutoring. In order to meet these challenges, this study uses a generative pre-training model to build an intelligent tutoring system for English teaching, aiming to innovate the English learning experience with the help of artificial intelligence technology and achieve personalized and efficient teaching guidance. The construction solution includes collecting data such as learners' English proficiency test scores, learning history, and self-reported learning preferences to create detailed learner profiles, integrating advanced generative pre-trained models such as GPT-based and fine-tuning with data related to English language teaching, and then automatically generating exercises based on learner profiles and dynamically adjusting the difficulty. The application of the system is reflected in the integration of natural language processing technology and generative models to provide immediate feedback after learners complete the exercises, such as analyzing the grammar, vocabulary use and coherence of English passages and pointing out mistakes, giving suggestions and explanations for corrections, as well as providing intelligent tutoring in the form of dialogues, such as examples, comparisons and related exercises to enhance understanding in response to learners' questions about grammar points. The experimental results show that compared with the traditional teaching mode, the use of this intelligent tutoring system increases the learners' progress in English listening, speaking, reading and writing by an average of 30 %, and the learning satisfaction increases by 40 %, especially in the improvement of oral expression and writing skills.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200232"},"PeriodicalIF":0.0,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143890966","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrating CNN and RANSAC for improved object recognition in industrial robotics","authors":"Yingding Xiao","doi":"10.1016/j.sasc.2025.200240","DOIUrl":"10.1016/j.sasc.2025.200240","url":null,"abstract":"<div><div>This research introduces a robotic grasping system that merges ORB (Oriented FAST and Rotated BRIEF) feature detection, VGG19 convolutional neural networks, and RANSAC (Random Sample Consensus) geometric verification to achieve high-precision object manipulation in unstructured environments. The framework synergizes ORB's efficient, rotation-invariant keypoints with deep semantic features extracted from intermediate layers of VGG19 enabling robust object recognition under occlusions and lighting variations. ORB detects scale-agnostic keypoints and generates binary descriptors, while VGG19’s hierarchical features provide contextual understanding of object geometry. These complementary features are fused into compact descriptors, combining ORB's 256-bit binary patterns with aggregated VGG19 layer outputs to balance accuracy and computational efficiency. RANSAC is then employed to eliminate mismatched features and estimate precise spatial alignments through iterative homography calculations, ensuring reliable mapping between detected objects and the robot's workspace. Experimental validation on industrial dataset trials demonstrates a 99 % grasp success rate, highlighting the system's ability to address challenges in dynamic, cluttered settings. By bridging deep learning's perceptual capabilities with geometric verification, this work advances autonomous robotic systems, offering a scalable solution for industrial automation that prioritizes precision and adaptability.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200240"},"PeriodicalIF":0.0,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143848560","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dance video action recognition algorithm based on improved hypergraph convolutional networks","authors":"Ni Zhen,&nbsp;Yiyi Jiang","doi":"10.1016/j.sasc.2025.200247","DOIUrl":"10.1016/j.sasc.2025.200247","url":null,"abstract":"<div><div>Dance as an art form is rich in movement and expression information. Accurately recognizing movements in dance videos is important for dance education, creation and performance. In view of this, the study takes the hypergraph convolutional network under deep learning as the framework basis, optimizes the performance by introducing the self-attention module and the topology module, constructs the temporal refinement channel and the channel refinement channel, and adds the spatio-temporal hypergraph convolutional network for channel fusion, and finally proposes a new video action recognition model. The experimental results show that the fastest iteration of this new model is 250 times, at which time the recognition accuracy is 95 %. The highest model P-value is 0.094, the highest R-value is 0.098, and the highest F1-value is 0.082. After the confusion test, the model shows &gt;90 % recognition accuracy. The accuracy, validity, and fluency scores were all above 90 after the dance category was rated by the judges. In summary, the study improves the hypergraph convolutional network and applies it to dance video action recognition with higher effectiveness and better recognition accuracy, the study aims to provide a more effective technical means for the development of dance education and performance field.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200247"},"PeriodicalIF":0.0,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143838543","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Practical exploration of English translation activity courses in universities under the background of artificial intelligence","authors":"Jiexia Dai","doi":"10.1016/j.sasc.2025.200249","DOIUrl":"10.1016/j.sasc.2025.200249","url":null,"abstract":"<div><div>In recent years, along with the increasing international exchange, cultivating English translation talents has become particularly important. This study aims to explore the application effect of artificial intelligence technology in college English translation activity courses and compare the differences between traditional teaching methods and AI-based methods. Through comparative analysis of the improvement of the translation ability of the two groups of students, the practical value and feasibility of AI technology in translation teaching is evaluated to provide reference and guidance for future translation education. This paper used the neural network machine translation model in artificial intelligence to solve the practical problems in implementing of college English translation activities. Intelligent translation through Recurrent Neural Network (RNN) can better leverage the role of artificial intelligence, which provides new methods for the sustainable development of translation activity classes in universities. Group A adopts the traditional English translation teaching method, mainly relying on teachers' explanations and students' practice. Course content includes interpretation of translation theory, text translation practice, and discussion and analysis based on translation cases. It mainly uses paper textbooks and reference books, as well as some online translation resources, but does not involve the application of artificial intelligence technology. Group B integrates artificial intelligence technology into English translation teaching method, using neural network translation model (such as RNN) to assist teaching. In addition to traditional translation theory and practice, the course also includes training and practice on the use of AI translation tools. Using AI translation software, an online translation platform and a specially developed AI-based translation practice platform, the platform is an artificial intelligence translation model integrated with RNN, designed to assist teaching and improve students' translation skills. The experimental results of this article indicate that the average translation scores of Group A and Group B students in the first 6 months of the experiment were 65.77 and 65.71, respectively. The scores of Group A and Group B students after the experiment for 6 months were 68.57 and 82.69, respectively. The results show that translation teaching in the context of artificial intelligence significantly improves the efficiency and accuracy of students' translation, and enhances the interaction and interest of learning. This finding shows that the AI-based technology in translation teaching is not only feasible and effective, but also can provide new ideas for the modernization and intelligent development of translation education, and improve students' practical translation ability and professional competitiveness.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200249"},"PeriodicalIF":0.0,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143877446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Stacked generalisation for improved prediction of joint shear in beam-column joints","authors":"Shruti Shekhar Palkar,&nbsp;T. Palanisamy","doi":"10.1016/j.sasc.2025.200246","DOIUrl":"10.1016/j.sasc.2025.200246","url":null,"abstract":"<div><div>Understanding shear and bond mechanisms becomes crucial in addressing the complexities and uncertainties inherent in designing beam-column joints, especially under seismic conditions. This study proposes employing machine learning as a viable approach for predicting joint shear strength. The study analyses a dataset containing 670 beam-column joints (312 interior and 358 exterior beam-column joints) statistical summaries featuring eight input parameters encompassing the joint's cross-sectional dimensions, reinforcement details, and material properties. Machine learning algorithms, including K-Nearest Neighbours, Support Vector Regressor, Decision Tree, Multilayer Perceptron, Random Forest, Extreme Gradient Boosting, Adaptive Neuro Fuzzy Inference System, Elastic Net, and Ridge, are fine-tuned using Optuna and evaluated based on performance metrics such as Root Mean Squared Error and R-squared. The study evaluates the performance of machine learning models for predicting joint shear strength in interior (IBCJ) and exterior (EBCJ) beam-column joints, emphasizing the effectiveness of ensemble techniques like stacking regressors. The Stacking Regressor consistently outperformed traditional models and design codes (CSA, ACI, AIJ, IS, GB, EN, and NZS), achieving RMSE values of 1.1407–1.2170 (R²: 0.8180–0.84) for IBCJ and RMSE of 1.02 (R²: 0.84) for EBCJ, compared to code errors exceeding 1.8. SHAP analysis revealed that concrete compressive strength (importance: 0.85 for IBCJ, 0.89 for EBCJ), column reinforcement percentage, and top reinforcement percentage were the most influential features. These findings highlight the potential of ML models to capture complex, non-linear structural behaviour more accurately than conventional methods. Keywords: Beam-Column Joint, Joint Shear Stress, Ensemble techniques, Feature Importance, Stacking Regressor, Algorithms, Machine learning</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200246"},"PeriodicalIF":0.0,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855606","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Analysis of community groups in large dynamic social network graphs through fuzzy computation","authors":"Ubaida Fatima ,&nbsp;Saman Hina ,&nbsp;Muhammad Wasif","doi":"10.1016/j.sasc.2025.200239","DOIUrl":"10.1016/j.sasc.2025.200239","url":null,"abstract":"<div><div>This study presents a fuzzy computing approach for dynamic community detection, termed Fuzzy Time-variant Community Groups (FTCG), utilizing fuzzy weighting techniques to track evolving network structures over time. The methodology was validated on a small 5-node graph and applied to large-scale datasets, including Amazon product networks, Bitcoin transactions, and Cellular Phone Network data. Two novel link-weighting techniques were introduced to enhance the detection of temporal community changes, while a Fuzzy Modularity measure was proposed to evaluate community quality. The impact of varying threshold values was analyzed, demonstrating how different thresholds influence community detection outcomes. Experimental results confirm the approach's effectiveness in capturing network dynamics, particularly in the Bitcoin and Cellular datasets, proving its robustness in Social Network Analysis (SNA) and its potential for informed decision-making in evolving systems.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200239"},"PeriodicalIF":0.0,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143838541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design of smart government IoT management system based on cloud platform","authors":"Wei Fu","doi":"10.1016/j.sasc.2025.200234","DOIUrl":"10.1016/j.sasc.2025.200234","url":null,"abstract":"<div><div>The underground pipe network and road lamps ensure the normal and orderly operation of urban life. The study starts with the intelligent control of urban SLs and the detection of underground pipeline water level and well cover displacement, using sensors in the urban IoT system to collect the environmental data of SL nodes and underground pipeline data. The DAG-SVM six-classification dimming classification model is trained in advance using data from SL nodes. According to the results of the classification model, the dimming ability of the street light is determined, and the illumination of SLs can be automatically changed through the IoT center. The underground pipeline data is used to monitor the underground water and well cover equipment, so that relevant problems can be solved in time to avoid problems such as untimely drainage when the urban flood season comes. To sum up, the research builds a IoTMS according to the CP. Through experimental analysis, the MS can save 29.3 % of energy consumption than manual control output. The water level detection error is &lt;3.5 %. It can realize the intellectualization of urban infrastructure, which has positive significances for urban energy conservation and flood season risk avoidance.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200234"},"PeriodicalIF":0.0,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855607","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Random-forest-based task pricing model and task-accomplished model for crowdsourced emergency information acquisition","authors":"Wenxiang Li,&nbsp;Shengqun Chen,&nbsp;Lijin Lin,&nbsp;Li Chen","doi":"10.1016/j.sasc.2025.200235","DOIUrl":"10.1016/j.sasc.2025.200235","url":null,"abstract":"<div><div>Over the past decade, crowdsourcing has emerged as a powerful tool in various scenarios and has led to an increasing need for crowdsourced emergency management. An important aspect of emergency management includes the acquisition of crowdsourced emergency information. Therefore, to study the emergency information acquisition, we propose a crowdsourced framework. During crowdsourcing, the public is recruited to work on collection of emergency information, such as photos and videos. Therefore, a considerable challenge in crowdsourced emergency information acquisition is to efficiently attract the public to engage in this work. A task price is a significant potential factor that influences public participation. Therefore, a random forest algorithm-based task pricing model and task-accomplished model are computed based on the task attributes and neighboring-workers attributes. In addition,the <em>making money by taking photos</em> dataset is used for a simulation of the proposed method in scikit-learn. Our simulation results demonstrate that the proposed method has an average reduction in Mean Squared Error (MSE) by 44.16 % for task pricing and an average increase in accuracy of 17.71 % for task-accomplished prediction compared to traditional regression models. It is shown that the proposed method has high accuracy and efficiency in crowdsourced emergency information acquisition. Moreover, the proposed method can provide valuable references in the future for emergency information acquisition strategy studies.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200235"},"PeriodicalIF":0.0,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143838542","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Combining state-of-the-art pre-trained deep learning models: A novel approach for Bangla Sign Language recognition using Max Voting Ensemble","authors":"Md. Humaun Kabir ,&nbsp;Abu Saleh Musa Miah ,&nbsp;Md. Hadiuzzaman ,&nbsp;Jungpil Shin","doi":"10.1016/j.sasc.2025.200230","DOIUrl":"10.1016/j.sasc.2025.200230","url":null,"abstract":"<div><div>Sign language is a crucial medium of communication for individuals with hearing impairments. Recently, many researchers have been working to develop an automatic sign language recognition system for English, Arabic, and Japanese languages to ease communication complexity between deaf and non-deaf communities. However, few systems have been developed for Bangla sign language (BdSL), and most of the existing ones may face difficulties in achieving satisfactory performance. While recent advances in deep learning have dramatically improved image classification tasks, including sign language recognition, ensemble methods offer a pathway for further enhancing BdSL identification accuracy. This study introduces a cutting-edge approach employing the Max Voting Ensemble technique for robust BdSL recognition. We have incorporated a range of cutting-edge pre-trained deep neural networks including Xception, InceptionV3, DenseNet121, ResNet50, and MobileNetV2. These models have been extensively trained on BdSL datasets, achieving individual accuracies ranging from 92.96% to 98.81%. Our method leverages the synergistic capabilities of these models by combining their complementary features to elevate classification performance further. In our approach, input images undergo preprocessing for model compatibility. The ensemble integrates the pre-trained models with their architectures and weights preserved. For each Bangla sign under examination, every model produces a prediction. These are subsequently aggregated using the Max Voting Ensemble technique to yield the final classification, with the majority-voted class serving as the conclusive prediction through comprehensive testing on a diverse dataset. Our ensemble outperformed individual models, attaining test accuracy of 96.62% and 99.92% using BdSL-38 and BDSL-49 datasets respectively, thus demonstrating superior BdSL recognition performance and reliability. We evaluated the effectiveness of our proposed method on the BdSL-38 and BDSL-49 datasets to ensure its generalizability. Our ensemble method delivers a robust, reliable and effective tool for the classification of BdSL. By utilizing the power of advanced deep neural networks, we aim to assist healthcare professionals in achieving timely and accurate diagnoses, ultimately reducing mortality rates and enhancing patient outcomes.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200230"},"PeriodicalIF":0.0,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143828856","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Research on automatic processing system of financial information in colleges and universities based on NLP-KG fusion algorithm","authors":"Jin Lei ,&nbsp;Mengke Wei ,&nbsp;Yiwen She ,&nbsp;Weixia Wang","doi":"10.1016/j.sasc.2025.200224","DOIUrl":"10.1016/j.sasc.2025.200224","url":null,"abstract":"<div><div>At a time when information technology is deeply embedded in the management of colleges and universities and the automatic demand for financial information processing is urgent, the automatic processing system of financial information in colleges and universities based on NLP and KG fusion algorithm has come into being, which can efficiently process massive data, improve work efficiency an scientific decision-making, but its data security and privacy protection are extremely critical. Traditional financial information processing in colleges and universities relies on manual entry and review, which is inefficient, error-prone and scattered, and early automation tools have problems such as insufficient semantic understanding and correlation analysis, and data silos. Financial data involves sensitive information such as fund receipts and expenditures, faculty and staff salaries, and student tuition, and relevant national laws and regulations also put forward strict requirements for its security and privacy protection. The existing mechanisms ensure security and privacy from multiple aspects: access control (using multi-factor identity authentication and role-based permission management), data encryption (SSL/TLS encryption protocol and AES algorithms are used for transmission and storage, respectively), data backup and recovery (regular backup, off-site storage, and recovery drills), audit and monitoring (detailed recording of operations and real-time monitoring of network traffic, etc.). The NLP-KG fusion algorithm further improves the system's data security and privacy protection capabilities through data semantic understanding to identify potential risks, intelligently adjust access permissions, and realize intelligent retrieval and analysis of encrypted data. In short, while the system brings opportunities, universities need to continuously improve data security and privacy protection mechanisms to cope with the complex cybersecurity environment and financial management needs.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200224"},"PeriodicalIF":0.0,"publicationDate":"2025-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143916898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}