Shanghai Municipal Human Resources and Social Security Bureau

Cross site scripting (XSS) attacks are still one of the key threats to the security of web applications. With the increasingly evolving attack methods, traditional detection methods have a high false positive rate of detection results. This article proposes a new model: a cascaded detection architecture that combines Long Short Term Memory (LSTM) networks with Large Language Models (LLM) for detecting XSS malicious code attacks. This method first uses Word2vec for code vectorization, turning the malicious code into vector data predictable by LSTM. Then, LSTM detects the vectorized malicious code and classifies the samples as malicious , benign , or suspicious. The malicious code samples classified as suspicious will be sent to LLM for detailed analysis again. The experimental results obtained show that compared to independent LSTM for malicious code detection, the cascaded method of LSTM and LLM has an accuracy rate of up to 97.8% and a 42% reduction in false positive rate, especially for some Base64 and URL encoded malicious code detection with higher accuracy. At the same time, to ensure detection speed, only 8.3% of the malicious code samples were extracted for detection.

Against the backdrop of the United States extreme financial sanctions on Russia, this paper conducts an in-depth analysis of the role Central Bank Digital Currencies (CBDCs) can play as a new type of financial infrastructure in responding to financial sanctions and reshaping the international monetary order under the dollar hegemony system. The study systematically examines the legal logic of financial sanctions under U.S. dollar dominance and the associated risks of weaponization, revealing the structural shortcomings and vulnerabilities of traditional cross-border payment systems such as SWIFT. It further elucidates the technical advantages and institutional potential of CBDCs in improving payment efficiency and reducing dependence on U.S. dollar channels.Building on this foundation, the paper focuses on the legal nature and technical characteristics of the digital renminbi (e-CNY), as well as the challenges it poses to the current monetary legal framework. Through an analysis of its applications in cross-border payments, settlement, and clearing scenarios, the study demonstrates how e-CNY can, in coordination with infrastructures such as CIPS and mBridge, form a composite system to advance renminbi internationalization. Finally, the paper systematically reviews the challenges faced by e-CNY in terms of cross-border interoperability, legal and regulatory coordination, geopolitical trust, and technological security. It proposes a series of legal and policy responses, including enhancing domestic legislation, actively participating in international rule-making, and strengthening technological security and system resilience. These recommendations aim to provide theoretical guidance and strategic options for China to safeguard financial sovereignty and establish a self-controlled international monetary circulation system amid an increasingly complex global financial environment.

With the diversification and personalization of mobile user demands, as well as the rapid development of distributed computing technology, leveraging mobile edge computing task offloading to provide users with convenient services has become a research focus. However, existing methods still face issues such as time consumption, high resource consumption, and data silos. Based on this perspective, we propose the federated Knowledge distillation based mobile Edge Computing task Offloading (KECO) method to achieve accurate task offloading and system energy saving. Through federated distillation learning, a teacher-student model is established, and complex and parameter-rich models are used as teachers to assist in the training of the student model. The teacher model transfers the knowledge and information it had learned to the student to enhance the generalization ability of the student. Finally, using the lightweight and flexible nature of the student model, it is deployed in a distributed system to implement the task offloading strategy in mobile edge computing. Experiments show that KECO performs well in large-scale task classification and allocation, and has efficient, reliable, and wide application potential.

Abstract This study introduces a novel hybrid forecasting model for coking coal prices, integrating complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and long short-term memory (LSTM) neural networks, enhanced by an improved fruit fly optimization algorithm (IFOA). The approach begins with CEEMDAN decomposing the coking coal price sequence into intrinsic mode functions (IMFs) and a residual component, effectively mitigating non-stationarity and nonlinearity. High- and low-frequency IMFs are differentiated using a single-sample T-test, with high-frequency components consolidated to minimize noise interference. Subsequently, the IFOA algorithm optimizes LSTM hyperparameters, boosting both generalization and prediction precision. Empirical validation, leveraging the Platts price index for four major imported coking coal varieties, demonstrates that the CEEMDAN-IFOA-LSTM model significantly outperforms a broad range of benchmarks, including ANN, IFOA-LSSVR, CEEMDAN-LSTM, LSTM, BiLSTM, TCN, IFOA-LSTM, CEEMDAN-FOA-LSTM, CEEMDAN-PSO-LSTM, and CEEMDAN-GA-LSTM, achieving reduced root mean square error (RMSE) and mean absolute percentage error (MAPE). The study concludes that this model adeptly addresses the challenges of nonlinear coupling and hyperparameter optimization, offering a reliable tool for coking coal price forecasting. Future research will aim to refine the model further to adapt to diverse market conditions and enhance forecasting accuracy.

Talent demand prediction is very important for national development strategy adjustment and social and economic development. Notably, this type of prediction is useful for industries to adjust their recruitment strategies and professionals to plan their careers. Therefore, automatically predicting the talent demand in society is necessary and may help job seekers to find a job. Many methods have been proposed to predict the talent demand. However, the accuracy of current methods needs to be improved. Therefore, in this paper, we propose a novel approach that uses a deep learning method to predict the talent demand in industry. First, we collect and build a talent demand data, build a corresponding dataset and preprocess the dataset. Second, we perform data modeling to prepare the input data for the deep learning model. Finally, we pass the modeled data to a specially designed deep neural network-based model to predict the quantity of talent. We also evaluate the proposed approach on the basis of the constructed dataset. The evaluation results suggest that the proposed method is accurate and outperforms state-of-the-art approaches. On average, it significantly reduces the mean absolute error by more than 26%, reduces the mean square error by more than 11%, and increases the squared correlation coefficient by more than 13%.

With the rapid computer vision technology, it plays an important role in modern agriculture, medical diagnosis, environmental monitoring and other fields. However, the computer vision technology mainly relies on RGB image information and ignores the spectral information of objects, which may lead to incomplete information or inaccurate recognition in some applications. Spectral sensors are capable of capturing spectral information of target objects, which provide richer data support for computer vision tasks; meanwhile, Convolutional Neural Networks and vision transformers have achieved remarkable results in image feature extraction and global information capture, and this paper proposes a research idea on complementary computer vision techniques based on spectral sensors, Convolutional Neural Networks and vision transformers, aiming to improve the performance and efficiency of computer vision tasks by combining the spectral information capture capability of spectral sensors with the advantages of Convolutional Neural Networks and vision transformers in image feature extraction and global information capture.

China’s labor market, being the largest globally, is facing challenges due to factors such as population aging and polarization of population movements, this study aims to examine the relationship between urban welfare and labor supply among urban residents in China, considering the mediating roles of social interaction and cynicism. OLS regression analysis using STATA16 software was conducted to test the correlations between urban welfare, social interaction, cynicism, and labor supply. The mediation effects were examined through KHB regression, demonstrating the specific roles of social interaction and cynicism as mediators. This study concludes that urban welfare has a significant negative impact on the labor supply of urban residents in China. Social interaction plays a positive mediating role, emphasizing its importance in the relationship between urban welfare and labor supply. Cynicism also plays a negative mediating role, albeit with a minor effect. When both social interaction and cynicism are considered, their joint effect is negative, with social interaction exerting a stronger mediation effect. Furthermore, the study highlights the gender and regional differences in the impact of urban welfare, social interaction, and cynicism on labor supply. These findings provide valuable insights for policymakers to enhance urban welfare programs and promote labor supply in China.

Major function-oriented zoning, a key spatial planning strategy in China, aims to coordinate resource endowments, socio-economic development, and subsequent planning initiatives. However, the existing framework for major function-oriented zoning relies predominantly on socio-economic statistical indicators at the regional level, often neglecting the critical role of carrying capacity. To address this limitation, we assessed both the current state and dynamic trends of the carrying capacity to identify risk and advantageous zones for major functions, with the objective of optimizing major function-oriented zoning in the Yangtze River Delta region, China. Our findings indicate that 47 counties are experiencing significant pressure under the current carrying capacity, while 57 counties exhibit a deteriorating trend in their capacity. Over half of the counties are categorized as having an overloaded carrying capacity. Based on this analysis, 66 counties have been designated as risk zones for major functions. Consequently, the optimization of major function-oriented zoning requires adjustments in 10 counties, incorporating the identified risk and advantageous zones to enhance spatial planning efficacy. This study proposes an enhanced methodological framework for major function-oriented zoning by fully integrating carrying capacity assessments, offering substantial support for territorial spatial planning in China. We believe that these improvements contribute significantly to more resilient and sustainable regional development strategies.