Systems and Soft Computing最新文献

{"title":"Analyzing the learning behavior patterns of business english learners using deep learning technology","authors":"Xiaohui Zeng","doi":"10.1016/j.sasc.2025.200259","DOIUrl":"10.1016/j.sasc.2025.200259","url":null,"abstract":"<div><div>This study employs deep learning technology to conduct a comprehensive analysis and prediction of learning behavior patterns among business English learners, making several innovative contributions. First, it applies a hybrid deep learning approach, integrating Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), to model both static and temporal aspects of learning behaviors. Second, the study identifies novel patterns, such as the strong correlation between high-frequency evening study sessions and improved academic performance, providing data-driven insights into effective learning strategies. Third, it demonstrates the feasibility of leveraging deep learning to dynamically adjust learning paths and offer real-time personalized learning recommendations, significantly enhancing learner engagement and outcomes. These findings lay the groundwork for integrating deep learning into intelligent education systems and highlight its potential to revolutionize personalized learning in the field of business English education.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200259"},"PeriodicalIF":0.0,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143932197","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":"BanglaLem: A Transformer-based Bangla Lemmatizer with an Enhanced Dataset","authors":"Md Fuadul Islam,&nbsp;Jakir Hasan,&nbsp;Md Ashikul Islam,&nbsp;Prato Dewan,&nbsp;M. Shahidur Rahman","doi":"10.1016/j.sasc.2025.200244","DOIUrl":"10.1016/j.sasc.2025.200244","url":null,"abstract":"<div><div>Lemmatization plays a crucial role in various natural language processing (NLP) tasks, such as information retrieval, sentiment analysis, text summarization, and text classification. However, Bangla lemmatization remains particularly challenging due to the language’s rich morphology and high inflectional complexity. Existing open-access datasets for Bangla lemmatization are limited in size, with the largest containing only 22353 unique inflected words, which constrains the effectiveness of data-driven neural models. To address this limitation, we introduce a novel dataset, BanglaLem, comprising 96040 frequently used inflected words. This dataset has been carefully curated and annotated through a rigorous selection process to enhance the accuracy and efficiency of Bangla lemmatization. Furthermore, we propose a transformer-based approach to lemmatization and evaluate the performance of various pre-trained and trained from-scratch transformer models on this dataset. Among these, the BanglaT5 model achieved the highest exact match accuracy of 94.42% on the test set. The BanglaLem dataset is publicly accessible via the following <span><span>link</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200244"},"PeriodicalIF":0.0,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143873954","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":"Multi-resource joint management strategy for 5 G network slicing based on POMDP","authors":"Yale Li","doi":"10.1016/j.sasc.2025.200242","DOIUrl":"10.1016/j.sasc.2025.200242","url":null,"abstract":"<div><div>Network slicing technology, as one of the key technologies of 5 G networks, can meet the communication needs of different scenarios by creating multiple virtual end-to-end networks on a unified infrastructure. However, how to effectively manage various resources in network slicing to improve service quality and resource utilization has become an urgent problem to be solved. Given this, to achieve joint optimization management such as computing resources and bandwidth resources, reduce network latency, and improve throughput and resource utilization, a network slicing resource management model based on partial observation Markov decision process is proposed. The model under consideration is predicated on partially observed Markov decision processes. Such processes are capable of perceiving changes in network topology and dynamically adjusting resource allocation to adapt to constantly changing network conditions. Furthermore, the model employs a hybrid heuristic value iterative algorithm to optimize computational efficiency, reduce network latency, improve throughput, and enhance resource utilization. After testing, the delay and throughput of the proposed resource management model increased with the increase in the number of service function chains. When the number of service function chains was 70, the delay was about 70 ms, lower than in other models. The throughput was about 250Mbit/s, higher than other models. The resource management model had 85 % and 81 % utilization rates of computing and bandwidth resources, respectively, which were better than other models. The above results indicate that the resource management model based on partially observed Markov decision processes can effectively reduce network latency, improve throughput and resource utilization, and has important application value for resource management of 5 G network slicing.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200242"},"PeriodicalIF":0.0,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143935032","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":"Forecasting and decision making of firm’s financial indicators based on the SSA-MLP-BPNN model","authors":"Xin Xu","doi":"10.1016/j.sasc.2025.200233","DOIUrl":"10.1016/j.sasc.2025.200233","url":null,"abstract":"<div><div>It is a complicated and important task to forecast and make decisions about financial indicators of listed enterprises, because accurate prediction can help enterprises better plan their financial strategy and business development. In recent years, with the development of artificial intelligence and machine learning technologies, more and more researchers begin to apply these technologies to the prediction and decision-making of enterprise financial indicators.In this paper, we develop a model combined with the Sparrow Search Algorithm(SSA), Multilayer Perceptron(MLP) and Back Propagation Neural Network(BPNN) (SSA-MLP-BPNN model) to study the prediction and decision-making of financial indicators of listed companies in China. By comparing the prediction results of SSA-MLP-BP model with other optimization algorithms, it is found that the SSA optimization algorithm performs superiorly in improving the performance of the MLP-BP model, and it is easier to find the global optimal solution, which improves the prediction accuracy of the model. The proposed algorithm can accelerate the convergence speed, leading to faster and more efficient training. Different optimization algorithms may perform differently on different datasets, so it is necessary to choose the appropriate optimization algorithm according to the specific situation. This study can provide reference for the prediction and decision-making of firm’s financial indicators.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200233"},"PeriodicalIF":0.0,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143886948","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":"HC-means clustering algorithm for precision marketing on e-commerce platforms","authors":"Dan Wu,&nbsp;Xin Liu","doi":"10.1016/j.sasc.2025.200236","DOIUrl":"10.1016/j.sasc.2025.200236","url":null,"abstract":"<div><div>With the rapid development of e-commerce industry, precision marketing has become a key means for enterprises to enhance competitiveness and profitability. However, traditional marketing methods often cannot accurately identify the characteristics of customers, leading to the waste of e-commerce resources. In this context, e-commerce enterprises urgently need a more accurate and efficient marketing method to meet the growing business needs. To this end, this study attempts to optimize the traditional K-means algorithm, and fundamentally improve the clustering effect in precision marketing by optimizing the selection of initial clustering centers and similarity measurement methods. Based on this, the research constructs an e-commerce marketing system based on HC-means algorithm to more accurately divide customer groups, identify high-value customers, potential customers and lost customers, and formulate differentiated marketing strategies for different groups. Experiments show that the average accuracy of HC-means algorithm in Glass database is 93.71, which is 15.48–15.79 higher than the highest accuracy of other two kinds of algorithms in the same kind of database. When the cluster number is 8, the Mahalanobis distance of HC-Means is reduced by 2.1 and 1.2 respectively compared with K-means and DBSCAN, which indicates that the clustering results are more reasonable in data distribution. When the cluster number is 3, more than half of the customers' consumption interval days are mainly concentrated between 8–12 days, and about 10 % of the customers make purchases every 2 days. These accurate customer behavior insights provide a strong basis for marketing strategy development. To sum up, the HC-Means system constructed by the research has achieved remarkable results in e-commerce precision marketing, greatly improving user satisfaction, and providing a valuable reference scheme for e-commerce enterprises to optimize marketing mode and achieve sustainable development.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200236"},"PeriodicalIF":0.0,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143878456","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 fusion generation algorithm of visual communication and product design based on AIGC technology","authors":"Guoying Chen,&nbsp;Xiaofeng Lan,&nbsp;Kai Liu,&nbsp;Can Cheng","doi":"10.1016/j.sasc.2025.200237","DOIUrl":"10.1016/j.sasc.2025.200237","url":null,"abstract":"<div><div>The current field of visual communication and product design is faced with some problems, such as low efficiency of creative inspiration acquisition, cumbersome design process and difficult to meet personalized needs. This paper analyzes the application of AIGC technology in visual communication, including the key role of AIGC generation model in design and its methods to improve design efficiency. The application of AIGC technology in product design and its change to the traditional design process are discussed, and the automatic design generation method based on AIGC is emphatically introduced. The design of the combination of straight face and inclined face improves the visual hierarchy, making the overall design perception score reach 593 points, which is 38 points higher than the previous design, indicating that the visual optimization effect is remarkable. In the design scheme generated by AIGC technology, the uniformity of color and material is improved by 4.66 %, and the success rate of systematic optimization design is 5.2 %, further improving the consistency and visual appeal of the design. In this experiment, the perceptual characterization model is validated using 28 indicators, providing a robust data foundation for design improvement. This paper makes an in-depth analysis of the requirements of fusion of visual communication and product design, and puts forward the basic framework of fusion generation algorithm and the method of dynamic fusion of visual communication and product design elements based on convolutional neural network. Finally, the effectiveness and advantages of the proposed algorithm are verified by experimental analysis.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200237"},"PeriodicalIF":0.0,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143883122","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":"Optimization design and application of mechanical characteristics in ergonomics of children's intelligent toys","authors":"Jinming Liu ,&nbsp;Yuanwu Shi ,&nbsp;Chengwei Gu ,&nbsp;Qingyi Li","doi":"10.1016/j.sasc.2025.200245","DOIUrl":"10.1016/j.sasc.2025.200245","url":null,"abstract":"<div><div>This study conducted a mechanical analysis on the ergonomic design of children's smart toys, with a focus on the stress conditions of the toys during use. By establishing an accurate mechanical model, we can delve into the stress distribution and deformation characteristics of toys in different usage scenarios. Using finite element analysis techniques, we simulated the dynamic response of toys under child interaction forces, revealing potential structural weaknesses and optimization space. In addition, the influence of material mechanical properties was also considered, and the most suitable material combination was selected accordingly. Although there are currently various types of smart toys on the market, only about 10 % of them have undergone ergonomic optimization. Using the experiential approach, commonly used smart toys are selected as research objects, and improved through the principles of ergonomics optimization design. The research results indicate that the ergonomic indicators of optimized smart toys designed specifically for children have significantly improved, with a pressure resistance of up to 120 kPa, effectively ensuring comfort and safety during use. In addition, an extended analysis of the empirical data obtained from this study provides strong support for further improving the design of smart toys that meet the needs of children. In summary, this comprehensive survey delves into the application of ergonomic principles to optimize the design process of children's smart toys, resulting in significant experiential results that can serve as a scientific basis for guiding future product improvements.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200245"},"PeriodicalIF":0.0,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143895586","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":"Influence of wearable biometric sensors on performance indicators of volleyball players","authors":"Guoqing Jia","doi":"10.1016/j.sasc.2025.200238","DOIUrl":"10.1016/j.sasc.2025.200238","url":null,"abstract":"<div><h3>Background</h3><div>Wearable sensors are now very common in sports, animation, and healthcare as well as in other fields. Wearable sensors allow sportsmen to monitor their performance, identify ailments, and provide important understanding of game dynamics. Particularly volleyball requires a variety of difficult motions, including digs and blocks, which are absolutely essential for the result of the game.</div></div><div><h3>Research Objectives</h3><div>The main goal of this work is to provide a wearable sensor-based technique for automating the detection and identification of volleyball-related events like digs and blocks. This seeks to replace the manual procedure whereby statisticians mentally note events during games.</div></div><div><h3>Methodology</h3><div>Data collecting for this work uses five Xsens MTw Awinda sensors. Two classification algorithms—K Nearest Neighbour (KNN) and Linear Discriminant Analysis (LDA)—are combined with two separate cross-valuation methods. We evaluate the KNN method using k values ranging from 1 to 10.</div></div><div><h3>Results</h3><div>With both cross-valuation techniques validating this conclusion, LDA beats KNN in terms of average accuracy. LDA gets an average accuracy of 99.56 % and 89.56 % correspondingly when contrasting classifications with four and 10 classes. With KNN (k = 5), for four and ten classes respectively the average accuracies are 66.08 % and 92.39 %.</div></div><div><h3>Conclusion</h3><div>This study shows how wearable sensors may be used to automatically detect and identify events connected to volleyball. The findings underline how better LDA is than KNN in reaching better average accuracy. These results can help to create more exact and effective techniques for monitoring and evaluating volleyball games.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200238"},"PeriodicalIF":0.0,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143851575","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":"Two lines of parallel translation of PMVS algorithm","authors":"Liying Fan","doi":"10.1016/j.sasc.2025.200241","DOIUrl":"10.1016/j.sasc.2025.200241","url":null,"abstract":"<div><div>Sparse 3D reconstruction by using the incremental motion recovery structure system. First, SIFT feature points in the English text sequence were extracted, and mismatches were removed by reverse screening method and RANSAC algorithm. According to the deficiency of PMVS algorithm in the reconstruction process, the corresponding improvement method is proposed. The PMVS algorithm was first used to obtain a rough quasi-English two-line parallel translation system, The projection matching points of the point cloud are obtained through the projection matrix, Then, the method based on the proximity point distance constraint, ZNCC stereo matching constraint and the pole line constraint is used for the regional diffusion of the matching points; Then use the template matching algorithm to obtain the corresponding matching block of the point cloud hole on two lines of parallel translated English text, The ZNCC stereo matching algorithm with the adaptive window size was used to obtain the matching points within the matching block, Finally, the spatial points corresponding to the matching points are obtained by sub-pixel interpolation and triangulation, Finally, a two-line parallel translation system is reconstructed. Classified the Chinese and English sentences into simple short sentences and complex long sentences. For simple short sentences, the rules-based and statistical methods are used to align the more complex long sentences, and then align the short sentences. In the phrase recognition stage, the Chinese-English bilingual \"marker words\" set is used to cut the Chinese-English sentences to obtain the \"marker words\" phrase. Then, the basic noun phrases were identified using a bilingual corpus-based approach. In the Temple dataset and Dino dataset, this paper proposes that the improved PMVS algorithm has 11.11 % and 10.64 % improvement in time efficiency compared to the original PMVS algorithm. The time used by the two algorithms in the first stage is given. According to the data in the table, for the data set Temple, the original algorithm takes 49 s, while the improved PMVS algorithm takes 85 s, which takes more time than the original algorithm.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200241"},"PeriodicalIF":0.0,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143890967","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":"Application and Research of Attention Mechanism Combined with Data Visualisation for Entrepreneurial Learning Course Recommendation System in Universities and Colleges","authors":"Chunhua Dong","doi":"10.1016/j.sasc.2025.200243","DOIUrl":"10.1016/j.sasc.2025.200243","url":null,"abstract":"<div><div>With the rise of entrepreneurship boom, the number of entrepreneurship courses in colleges and universities is increasing. However, the traditional course recommendation system is often lacking in individuation and cannot adapt to the dynamic changes of students' needs. Therefore, the study proposes an innovative converged recommendation system that combines Attention Mechanism (AM) with Data Visualization (DV) techniques to enhance personalized recommendation capabilities for entrepreneurial learning courses. By analyzing students' interests and needs in real time, this method uses attention mechanism to dynamically adjust recommended content, while using data visualization technology to visually display course characteristics, so as to improve students' participation and learning effect. Extensive performance testing on the Enlec dataset showed that the fusion system significantly outperformed traditional methods in both recommendation accuracy and coverage, with an overall recommendation accuracy of 99.4 %. In the results of the recommendation test for 685 students, the highest course selection rates for the four systems were 74 %, 71 %, 68 % and 63 %, respectively, while the recommendation effectiveness of the integrated entrepreneurship course reached 98.5 %. The results confirm the effectiveness and robustness of the proposed method in practical application. The final results show that the proposed system not only improves the course selection rate of students, but also significantly enhances their interest in entrepreneurial learning courses, providing an effective solution for personalized learning in higher education.</div></div>","PeriodicalId":101205,"journal":{"name":"Systems and Soft Computing","volume":"7 ","pages":"Article 200243"},"PeriodicalIF":0.0,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143886947","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}