A large amount of evidence suggests that companies with good ESG performance have a lower risk of collapse and stakeholder risk, thereby diluting systematic risk. However, compared to fundamental financial indicators, ESG as a non-financial indicator has not yet reached a consistent conclusion on its mechanism, dynamic variability, and heterogeneity of impact on stock systematic risk. Based on this, this paper selects data from January 2009 to November 2023 in the A-share market for empirical research. Based on the conditional CAPM model, the systematic risk $\beta$ is dynamically characterized as a linear function of ESG performance (non-financial characteristics) and company fundamental characteristics (financial characteristics). Then, the MCMC Bayesian estimation method is used to obtain time-varying estimates of $\beta$ for results analysis. The research results are as follows: First, there is a negative correlation between ESG performance and stock systematic risk, which has become increasingly strong and significant in recent years. Second, the impact of ESG performance on stock systematic risk shows heterogeneity across industries. For industries that are more affected by energy or national policies, good ESG performance helps to reduce systematic risk. Third, although ESG performance can affect stock systematic risk, investors respond less to ESG risk than to fundamental risk, leading to asymmetric investor reactions. Therefore, ESG risk can be considered a secondary risk, and its impact on systematic risk is moderated by fundamental characteristics such as market value and book-to-market ratio.

The "fundamentals" that determine expected stock returns consist of three components: The expected return component that is rationally compensated for systemic risk, the cash flow news generated by changing expectations of future cash flows, and the discount rate news generated by changing expectations of rational future discount rates. Of these, cash flow news that are not driven by fundamentals are the component most likely to reverse in the short term. In this paper, for the first time, we use analyst forecast revisions to measure cash flow news and construct a short-term reversal strategy based on residual returns to study the short-term reversal effect in the Chinese Stock Market. The empirical results find that the strategy generates higher risk-factor-adjusted returns compared to the standard reversal strategy. Further, the methodological test based on isolating cash flow news finds that liquidity shocks rather than investor sentiment can explain the short-term return reversal phenomenon in the Chinese Stock Market from both the long and short leg of the portfolio. This paper examines the short-term reversal strategy of A-shares from a new perspective of residual returns, which not only enriches the research in the area of financial anomalies in asset pricing, but also provides international evidence on understanding the theory of short-term return reversal in emerging markets, with China as an example.

To address the challenges of passenger difficulty in hailing rides and insufficient vehicle supply during peak periods, this paper is based on an actual scenario in which ride-hailing platforms implement both surge pricing and subsidy strategies during peak periods, taking into account the heterogeneity of driver supply prices. By incorporating the reference point dependence theory of the driver decision-making and aiming to maximize platform expected profits, a Stackelberg game model is established to analyze the interaction between the platform and drivers. Kuhn-Tucker theorem is applied to determine the optimal service price and subsidy level for the platform during peak periods. The results reveal that the expected profit of platforms during the peak period showed an "inverted U-shape" trajectory as subsidies increased. When the subsidy level is within a critical range, the expected profits of the platform initially increase and then decrease with increasing service prices. The maximum expected profits of the platform are achieved when the critical value (peak point) is reached. By adopting appropriate pricing and subsidy strategies, ride-hailing platforms during peak period can effectively reduce the risk of driver attrition, adjust supply-demand relationships, and enhance the benefits for both the platform and drivers.

With the rapid development of the Internet, people have become accustomed to sharing hobbies, obtaining information and discussing common hot topics on the Internet, studying the law of public opinion communication in multi-layer social networks is beneficial to public opinion analysis and public opinion governance. Based on the traditional SEIR epidemic model, this paper considers the influence of node importance on propagation probability, and introduces dynamic parameters, and constructs a single-layer network public opinion propagation model. At the same time, considering the impact of different time steps and degree correlations on public opinion propagation, a multi-layer network cross propagation public opinion propagation model is proposed. In this paper, theoretical verification and experimental analysis are carried out on various communication performances and laws of multi-layer network public opinion communication model. Experiments show that time step and degree correlation have a significant impact on public opinion communication. Finally, some public opinion governance mechanisms and public opinion response measures are put forward, which can help the government and relevant administrative departments to improve the efficiency of public opinion management, ensure rapid response in public opinion events, and reduce potential negative effects, and this is of great significance.

A variety of hot topics lists released by social media platforms serve as a convergence and showcase for hot topic information, which provides significant insights toward our understanding of current popular discussions. However, due to vocabulary sparsity and short text length in hot list texts, traditional LDA and neural network-based topic mining models face poor performance in topic aggregation. To address these challenges, the paper proposes a topic modeling framework enhanced by a large language model—STAB, which combines the generative capabilities of large language models for text data with the excellent performance of document embeddings in topic modeling, enabling the extraction of meaningful topics from short text datasets. Experimental results on multiple datasets show that our framework outperforms existing topic modeling methods in terms of general objective evaluation metrics and applications in downstream tasks.

The development of rural revitalization industry is the basis for stimulating the vitality of rural areas and the premise for solving all problems in rural areas. Based on the development status of the “new community factory” in the Qinba Mountains of Shaanxi Province, this paper identifies the important participants such as community factories, local governments and leading enterprises, and constructs a tripartite evolutionary game model considering factors such as cost, subsidy, investment and income, and then discusses how each subject makes strategic choices in the process of rural revitalization industry development, and conducts sensitivity analysis and numerical simulation. The results show that, on the one hand, the evolutionary stabilization strategy is affected by the local government subsidy, and the appropriate subsidy is conducive to the joint participation of the three parties. The investment of leading enterprises in production enterprises should be within a relatively reasonable range, and at the same time, it is necessary to increase the investment of all parties in society to the local government; On the other hand, for community factories and leading enterprises, reducing costs and increasing profits will be more conducive to promoting the development of rural revitalization industries such as “new community factory”.

The setting of the lowest sampling rate of the subway white list security inspection channel is a difficult problem to balance safety and efficiency of the white list channel. This paper summarizes and condenses three kinds of white list credit supervision modes of subway security inspection: Government supervision, professional institution supervision and meta-regulation supervision, and then constructs the lowest sampling rate model of white list channel considering credit preference under the two scenarios of “credit” constraint and “credit + convenience” constraint. The main findings are as follows: (i) When the influence of the convenience income of passenger transport is not considered, the credit supervision mode with the largest credit gain-loss distance is dominant, and the meta-regulation supervision mode is generally better than the professional institution supervision mode; (ii) The higher the mutual recognition of the results of passenger credit supervision among regulatory agencies, the lower the lowest sampling rate of the white list channel; (iii) The credit gain-loss distance is positively related to the degree of credit preference of white list passengers, and the two types of credit preference are superposed; (iv) Credit constraints and convenience constraints are complementary. At the same time, when the mutual recognition rate of government departments and professional institutions for credit supervision results is low, or the credit preference of white list passengers for both types of credit is low, the sensitivity of passenger convenience to the lowest sampling rate of white list channels will be significantly enhanced; (v) When the proportion of white list passengers is too high or the number of white list channels is too small, the convenience constraint effect of white list passengers will be significantly reduced, and even the credit constraint effect will be eroded. Finally, the lowest sampling rate of white list channels under the “credit + convenience” constraint will be higher than the value under the “credit” constraint.

The anomaly detection has been a widely concerned topic of great application value and research importance for a long time. There are many machine learning algorithms dealing with the anomaly detection problem without clear statistical guarantee on the degree of false discovery. We propose a general framework for anomaly detection based on conformal inference that enables online false discovery rate control and does not rely on any model or distribution assumptions. The proposed procedure can incorporate different machine learning algorithms and online multiple hypothesis testing algorithms, thus providing a flexible and versatile approach for anomaly detection. We verify the effectiveness of the proposed procedure on simulated data and apply it to Server Machine Dataset to detect anomalies.

In the numerous terrorist attacks that have occurred worldwide, various terrorist organizations have shown a trend of collaborative cooperation, posing significant challenges to international counter-terrorism efforts. Based on the global terrorism database (GTD), this study constructs a terrorist organization cooperation evolution network from 121,074 terrorist attacks that occurred globally from 2001 to 2018 and conducts a time-series topological structure analysis. Based on the characteristics of terrorist organization cooperation, the network is divided into time slices of three years each to model the flow patterns of terrorist communities at multiple scales. The analysis shows that the robustness of the terrorist organization cooperation network has been continually strengthening over time, which is necessary to develop corresponding strategies to disrupt it. Focusing on the largest connected sub-network within the terrorist cooperation network, whose influence is continuously expanding, this study proposes a community structure-based neighborhood centrality index (CSNC) to measure the importance of nodes in the largest connected component. Experimental results demonstrate that the network disruption strategy based on CSNC, in the process of disintegrating the terrorist cooperation network from 2001 to 2018, achieved a 16.45% maximum reduction in the R value compared to benchmark strategies, proving that the CSNC-based disruption strategy can more effectively dismantle terrorist cooperation networks.

Consistency is one of the key and challenging issues in decision-making processes, which affects the validity and accuracy of preference relations and decision results. When the decision makers' preference is expressed by the hesitant fuzzy linguistic term set, in order to obtain the decision information in the linguistic decision environment effectively, this paper studies the consistency ascertaining, inconsistency repairing and weight derivation for hesitant fuzzy linguistic preference relations. First, two new definitions of additive consistency for hesitant fuzzy linguistic preference relations are proposed: Completely additive consistency and weakly additive consistency. Based on the proposed definitions of additive consistency, some linear programming models and 0-1 mixed programming models are provided to ascertain the consistency. In addition, for the incomplete hesitant fuzzy linguistic preference relations, some methods are developed to estimate the missing values and ascertain the consistency. When the hesitant fuzzy linguistic preference relations are inconsistent, a repairing method is developed, then the priority weights are calculated based on the consistency. Finally, some specific examples of decision problems and comparative analysis are provided to show the effectiveness and rationality of the models.

For the sales system composed of a manufacturer and a retailer, this paper constructs three decision models that the manufacturer does not introduce the live streaming channel, the manufacturer introduces the store live streaming channel and introduces the anchor live streaming channel, and uses the game theory method to explore the introduction strategy of the manufacturer's live streaming channel and the pricing strategy of multi-channel supply chain under the introduction of the live streaming channel. The results show that after the introduction of the live streaming channel, the manufacturer's revenue increase, the price and sales in the direct sales channel decrease, and the price decreases while the sales increase in the offline retail channel. When the cost of the store live streaming channel is low (high), the introduction of the store (anchor) live streaming channel is conducive to the “low price” strategy of the live streaming channel. When the cost of the store live streaming channel is low and the professional capability of the anchor is weak, the introduction of the store live streaming channel is conducive to the market “penetration” strategy of the live streaming channel, under other conditions, the introduction of the anchor live streaming channel is conducive to the market “penetration” strategy of the live streaming channel. Finally, the price is not necessarily cheap in the live streaming channel.

Considering a second-order nonlinear multi-agent systems with time-delayed, this paper investigates the leader-following consensus problem for this system under a fixed directed topology. In order to save the communication and computation resources of the system, a distributed consensus control protocol based on event triggering mechanism is presented. A corresponding event triggering condition is designed for each agent, and the agent updates the controller only when the triggering condition is satisfied, the frequency of updating information among the agents is effectively reduced. Some sufficient conditions for the multi-agent system with time-delayed to achieve leader-following consensus are given using graph theory, matrix theory and Lyapunov stability theory, and the Zeno behavior of the system is strictly excluded. Finally, the feasibility and effectiveness of the proposed control protocol are verified by simulation results.

Salary incentives are an important means for enterprises to stimulate employees' work passion, creative potential and improve corporate performance. On the one hand, the internal pay gap has a positive motivating effect and promotes the effort of employees. On the other hand, it can also lead to a sense of unfairness and cause some employees to feel “flat”. This paper constructs a mathematical model including core executives, non-core executives, and ordinary employees to analyze the impact of internal salary gaps on corporate performance, and conducts empirical research using data from privately-owned listed companies in Shanghai and Shenzhen from 2008 to 2020. Both theoretical and empirical results show that the relationship between pay gap within management, executive-employee pay gap, the degree of compensation incentives of non-core executives and corporate performance all show an inverted U shape. Further empirical research shows that non-core executive-employee pay gap has the strongest effect on corporate performance, while core executive-employee pay gap has the smallest effect on corporate performance. This research shows that non-core executive-employee pay gap is the most important compensation relationship within the company and core executive-employee pay gap is of least importance. In addition, in the salary incentive design of private enterprises, the “constraint” of operating profit is greater than the “constraint” of operating income, which reflects that private enterprises pay attention to seizing the key points.

To analyze the cooperation behavior from the perspective of doctors and patients in the smart medical information platform, this paper constructs a random evolutionary game model and introduces random interference factors to represent the information uncertainty in the information platform. Afterward, this paper transforms the stochastic evolutionary game model of doctors and patients into a cusp catastrophe model through the limit probability density function, proving that the sudden change mechanism is implied in the behavior evolution of doctors and patients. Finally, the simulation method is used to analyze the evolution and sudden change mechanism of the behavior selection of doctors and patients in the diagnostic information platform and the interactive information platform. This paper finds that: In the smart medical information platform, the behavior selection of doctors and patients may change drastically due to factors such as perceived value and random interference; When the initial cooperation probability between doctors and patients is high if the security and reliability of information platform are poor, doctors and patients tend not to cooperate; When the initial cooperation probability is low if the information platform security and reliability are good, doctors and patients tend to cooperate; When the initial cooperation probability is at a moderate level, the behaviors of doctors and patients in the information platform are more susceptible to the additional benefits of non-cooperation, sudden changes of behavior selection are more probable. This paper analyzes the cooperation mechanism between doctors and patients in the smart medical information platform by combining evolutionary game theory and cusp catastrophe theory. It explains the reasons for sudden discontinuous changes in the behavior state of doctors and patients. The conclusion provides enlightenment for the research on doctor-patient cooperative behavior in the information platform and the development of the smart medical platform.

Promoting new energy vehicles is an important measure to effectively reduce the carbon footprint of the transportation system. Based on the consumer utility theory, we construct a decision model for both charging and switching modes of new energy vehicles. This model enables a comparative analysis of pricing and promotion mechanisms for the two service modes. Additionally, we examine the impacts of battery production cost, switching model technology level, driving range, and energy prices on the promotion of both modes. Our main results are as follows: 1) Compared to the charging mode, the switching mode effectively alleviates consumer charging anxiety, albeit at the expense of a premium for switching services; 2) The production cost of power battery and the level of power change technology are important dimensions that affect the adoption of the two service modes; 3) There exists a divergence in the influence of battery production costs and driving range on the promotion of the two modes; 4) Decreasing battery production costs prove more beneficial for the promotion of the switching mode, while an extended driving range is more advantageous for the promotion of the charging mode. Changes in electricity or fuel prices exert similar effects on the promotion of both modes, accelerating their application by establishing operational cost advantages for new energy vehicles.

In the context of the digital economy, the monopolistic behavior of platform enterprises and the design of antitrust mechanisms have become hot and difficult problems in current research. To solve this problem, a tripartite evolutionary game model among government, platform enterprises and the public was established under both static and dynamic reward & punishment mechanisms. In terms of model analysis, the possibility of mixed strategy Nash equilibrium (MNE) as an evolutionary stability strategy (ESS) under the two reward & punishment mechanisms and its system evolution characteristics were mainly discussed. It was proved that MNE under dynamic reward & punishment mechanisms may be the system ESS, and confirmed through system simulation. The simulation results indicate that under the static reward & punishment mechanism, all parties in the game will exhibit periodic strategy selection patterns, while under the dynamic reward & punishment mechanism, the system gradually stabilizes to MNE. This indicates that the existence of the dynamic reward & punishment mechanism is indeed a stability improvement compared to the static reward & punishment mechanism. Finally, it is suggested that government should develop dynamic reward & punishment mechanism, while increasing the intensity of punishment on platform enterprises and gradually reducing the intensity of rewards for the public. This approach can significantly increase the probability of compliance operation of platform enterprises and improve the expected utility of government and the public.

Meta-analysis is a statistical method that systematically integrates, analyzes and synthesizes the results of multiple independent studies to reach more accurate and comprehensive conclusions. To solve model uncertainty in the prediction for meta-analysis, an optimal model averaging prediction method is proposed based on Mallows criterion, and the optimality of Mallows model averaging(MMA) estimator under square loss is discussed. Finally, simulation studies are conducted to evaluate and compare the performance of MMA, Jackknife model average(JMA), S-AIC and S-BIC model average estimation under information criteria, and all methods are applied to analyze the data set of BCG vaccine for illustration. The results show that the MMA estimation is superior to other model average estimations in prediction regardless of whether the variance and sample size are large or small.

Road condition detection is a core task in intelligent driving, including height limit detection tasks. Considering that the research related to height limit detection in the academic community is not yet mature, we have conducted research on height limit detection methods and proposed a height limit detection network based on key points and multi-frame image feature fusion. By adopting key points in the height limit detection task, unnecessary predictions are reduced and detection efficiency is improved. By introducing a convolutional gated recurrent unit (ConvGRU) to model multiple images and learn the contextual relationship between multiple images, improve recall rate, and reduce missed detection rate. The spatial particulars feature (SPF) module is proposed, which strengthens the multi-scale feature fusion in the decoding layer. In order to improve the accuracy of the model, the coordinate attention mechanism is introduced, and the target detection area is further paid attention to. According to the experimental results, this network can not only complete the height limit detection task well, but also balance the precision and recall rate better, with higher F1 values and fewer parameters compared with other advanced networks such as BiSeNet, PINet, PSPNet, etc; At the same time, in the task of lane line detection, it also performs excellently in terms of accuracy and missed detection rate, further proving the effectiveness of the network.

Due to “the curse of dimensionality”, both parametric and non-parametric high-dimensional tests are exposed to the issue of low power. Currently, there are two approaches to enhance the power of high-dimensional tests: 1) Add an indicative function to the test statistic, and use the marginal information to promote the power of high-dimensional tests, called as power enhancement. 2) Apply the sample splitting technique for dimensionality reduction of variables to improve the power of high-dimensional tests, named dimensionality reduction method. Based on these two ideas, this paper proposes the hypothesis tests via power enhancement and dimensionality reduction for high-dimensional means, regression coefficients and independence respectively. Numerical results demonstrate that the power enhancement method can obtain high power under both sparse and dense hypotheses. But the test level depends on the selection of the original test statistic and the threshold parameter. The dimensionality reduction method can control the significance level pretty well without considering the threshold parameter selection. Under the sparse hypothesis, the dimensionality reduction method possesses high power, but it performs lower than the power enhancement under the dense hypothesis.

Zeroing neural network (ZNN) has been widely used to solve time-varying problems since it was proposed because of its fast convergence speed and ability to resist external noise interference. However, the convergence speed and anti-interference ability of the existing zeroing neural network models are still not satisfactory. Therefore, to further improve the performance of ZNN, a new fixed-time convergent activation function (FTCAF) is designed in this paper. Then, a fixed-time convergent zeroing neural network (FTCZNN) model is established based on the proposed activation function and this model is applied to solve dynamic Sylvester equation (DSE). Theoretical analysis proves that the FTCZNN model has a fixed time convergence upper limit and strong anti-interference ability. In addition, numerical simulation results also demonstrate the superior performance of the FTCZNN model. Finally, FTCZNN model is used to realize the trajectory tracking experiment of the robot manipulator. The experimental results once again prove that the FTCZNN model has fast convergence speed and strong anti-interference ability, and its practical application ability is also verified.

Due to the non-subadditivity property of value-at-risk (VaR) when measuring tail loss risk, we propose a risk parity investment portfolio model based on conditional value-at-risk (CVaR) and provide a numerical calculation method for implementing the investment portfolio strategy. Using Sharpe ratio, maximum drawdown, and Calmar ratio as performance evaluation indicators, the risk parity investment strategy based on CVaR is compared with common investment portfolio strategies. Numerical experimental results indicate that the comprehensive performance of the risk parity strategy is more robust than the equal-weight investment portfolio strategy, the maximum Sharpe ratio investment strategy, and the global minimum variance investment strategy. Among the three risk parity strategies, the CVaR-based risk parity investment strategy has advantages in risk control, significantly improving both return and risk diversification effects. The robustness test results also suggest that the CVaR-based risk parity investment strategy can maintain stability and effectiveness in different situations.

This paper studies boarding probabilities of metro passengers at platforms. Under several reasonable assumptions, we prove a stochastic queueing model for passengers' boarding probabilities, which extends the deterministic queueing model (first-come-first-serve principle) in the research of Grube, et al. (2011) and Mo, et al. (2020). We present a simplified version of the stochastic queueing model that reduces to the deterministic queueing model when the parameter of the version is set to be zero. Based on the stochastic model, we construct a passenger flow simulation on a typical route. With the train schedule, passengers' tap-in times, and walking time distributions being inputs, the simulation yields each passenger's movement and tap-out time as outputs. Combining real data and simulation outputs, we provide a parameter calibration method. Real data analysis on Changping line of Beijing Metro illustrates advantages of the proposed stochastic model over the existing deterministic model.

Cylindrical algebraic decomposition is a basic tool in semi-algebraic system solving and real quantifier elimination. In the actual solving process, the choice of a variable ordering may have a significant impact on the efficiency of cylindrical algebraic decomposition. At present, the existing heuristic or machine learning ordering selection methods are basically based on the implicit assumption that the support set of a polynomial system is the determinant for affecting the variable orderings. In this paper, we first test this hypothesis by designing an experiment with the support set fixed but the coefficients varying. The experimentation shows that the support set is indeed an important factor, though not the only factor, determining the optimal variable ordering. Aiming at selecting the optimal ordering for computing cylindrical algebraic decompositions for systems with the same support set but different coefficients, this paper designs an ordering selection scheme via reinforcement learning. The experimentation on four variables shows that this scheme can surpass the accuracy limit of existing methods on selecting the optimal variable ordering that rely solely on the support set. In addition, experiments on systems owning up to 20 trillion of possible orderings show that the scheme is much more efficient than traditional heuristic methods. In contrast to the existing supervised learning methods for selecting the variable ordering of a few variables, this reinforcement learning scheme overcomes the difficulty of obtaining high-quality labeled data when the number of variables increases, which may lead to the combinatorial explosion of the number of variable orderings.

In group decision-making under the social network environment, the weight of experts in the group and the trust relationship between experts are key factors that affect consensus reaching. However, in many studies, the trust relationship remains unchanged and the expert weight is only determined by the trust relationship. Therefore, this paper innovatively proposes a group consensus decision-making method that considers the social influence and trust evolution of experts, effectively promoting the reaching of group consensus. Firstly, the incomplete social trust matrix is transformed into a complete social trust matrix using trust propagation and aggregation methods. Then, the social influence of experts is obtained based on their additive preference relationship and social trust matrix, and the weights of each expert are obtained. Subsequently, a trust evolution model is established based on whether the optimal solution of each expert has been adopted and the difference between the ranking vectors of each expert's solution and the group's solution. Based on the trust evolution model, a consensus reaching process considering trust evolution is proposed. By using simulation methods, the weight coefficients of various indicators in social influence are calculated, and the feasibility of the proposed consensus reaching method is verified to demonstrate the rationality and effectiveness of the proposed model. Finally, a numerical example is presented to illustrate the detailed solution process of the method proposed in this paper, further demonstrating the feasibility and effectiveness of the model.

To improve the accuracy of unmanned underwater vehicle (UUV) state estimation of non-cooperative targets, an axial attention-based target state estimation method is proposed in this paper. The state estimation mechanism of the UUV non-cooperative target based on sonar observation is analyzed. The non-Markov state-space model of the problem is transformed into a first-order Markov state-space model with memory, and a recursive filtering model is constructed. Aiming at the unreliability of forward-looking sonar observation and the unpredictability of target motion, a multi-step prediction network based on transformer is proposed to describe the complex motion process of non-cooperative target relative to sonar under nonlinear observation. Aiming at the instability of observation and the unpredictability of posterior distribution, based on the Monte Carlo approximate inference principle, the multi-step prediction network is used to map the particles in the target measurement state space to the target prediction state space, and a non-cooperative target state estimation algorithm based on the axial attention is constructed. The simulation results show that the adaptability and robustness of the proposed method to uncertain inputs.

The overwhelming selection of movies on the market nowadays makes it difficult for users to make a decision. An efficient movie recommendation system plays a significant role in improving user experience and the market competitiveness of movie service providers. The challenge lies in how to integrate multiple data sources for personalized recommendations while balancing algorithm accuracy and diversity. Research on this issue is of great theoretical and practical significance. User portraits can depict rich user characteristics from multiple dimensions, helping us to better understand user interests and behaviors. Meanwhile, link prediction offers special benefits when modeling from a network topology standpoint. The integration of them provides a possibility to solve the above issues. Therefore, this study proposes a novel user portrait and link prediction-based personalized recommendation algorithm called UPLPR. The algorithm is designed under the background of movie recommendation. It distinguishes between the interest similarity that reflects in user behavior and genre domain. By abstracting user portraits from multiple data sources and integrating them into the link prediction process as external information of the network, the accuracy of the algorithm can be improved. Furthermore, from the perspective of scarcity, the algorithm improves the calculation of interest similarity between users in bipartite graph projection and evaluates the promoting or inhibiting effect of links in the recommendation process. Such consideration improves the novelty and personalization of the recommendation and mitigates the popularity bias problem to some extent. Finally, experiments were conducted on two MovieLens datasets to verify the proposed recommendation algorithm. Results show that compared with representative algorithms, the algorithm proposed in this paper not only achieved significant performance in accuracy but also demonstrated obvious advantages in diversity-related indicators. Additionally, the abstracted user portraits can help recommendation platforms understand their user base, thereby formulating more scientific marketing and management strategies.

The main purpose of mediation analysis is to study whether the independent variable affects the dependent variable through the mediator variable. The mediation analysis method has been widely studied and applied in the fields of social science and biomedicine, and has become the mainstream research method in these fields. Whether it is a single mediator variable or multiple mediator variables, many scholars have conducted extensive research, but it is worth noting that these studies on mediation analysis seldom use prior information. In the case of high dimensions, the prior information extracted from the existing research can improve the performance of estimation and testing. In this paper, we propose a mediator variable selection method and a hypothesis testing method based on prior information. Theoretical and numerical analysis results show that the method proposed in this paper outperforms the existing methods when the prior information quality is good, and shows robustness when the prior information quality is moderate, and the performance is not much different from the existing methods. Finally, the real data example further illustrates the satisfactory performance of the proposed method.

In order to investigate the impact of significant economic or political events on the dependence of financial markets, we construct the factorial hidden Markov Copula model (FHM-Copula) that allows the coefficients of dependence to follow a regime-switching process in high-dimensional state space. The FHM-Copula model is able to capture external shocks of varying magnitude, direction, duration, and short or long-term from significant events to the dependence. In the empirical study, we analyze the dynamic dependence between the stock markets of China and other BRICS countries by adopting the FHM-Copula approach. Our findings indicate that the FHM-Copula model can effectively identify the external shocks caused by significant events such as the subprime crisis, the European debt crisis, the Chinese stock market crash, China's taking over the BRICS presidency and the COVID-19 epidemic on the dependence between the stock markets of China and other BRICS countries. Our works not only provide a theoretical analysis framework based on the information shock perspective for the study of dynamic dependence among financial variables, but also provide a reference for investors and government regulators in investment decisions and risk management.

During the 14th Five Year Plan period, the reform of green fiscal and tax policies was proposed to strengthen the regulatory and guiding role of incentive tax policies, such as value-added tax. This paper is based on a quasi natural experiment of China's tax system reform from business tax to value-added tax. By constructing a multi-time point double difference model, we analyze panel data of 1805 A-share listed companies and examined the impact mechanism of tax policy incentives on the green innovation of enterprises. The research finds that the replacement of business tax with value-added tax significantly enhances green innovation in enterprises, mainly reflected in substantive innovation rather than strategic innovation. Research on the mechanism of action indicates that the reform from business tax to value-added tax indirectly promotes green innovation in enterprises through the effects of division of labor and the tax burden. Heterogeneity studies have shown that the incentive effect of replacing business tax with value-added tax on green innovation is more prominent in non-state-owned enterprises and manufacturing enterprises. This paper is an important supplement to the research on existing tax policy reform and micro-enterprise behavior, providing an important basis for promoting green development through financial and tax policy reform in the future.

This paper studies the stabilization problem of heterogeneous multi-agent systems composed of the first-order and second-order dynamic agents in a signed directed graph. By utilizing the knowledge of Laplacian matrix and graph theory, corresponding protocols are designed for the second-order and first-order dynamic agents, respectively. Based on the layering theory proposed in this paper, independent strongly connected components (SBiSCC) of structural equilibrium are utilized to design control parameters. The necessary and sufficient conditions for achieving stabilization of the first-order and second-order heterogeneous systems in communication topology are given. Finally, the paper provides several simulation verification theoretical results.

This paper deals with a second-order conic programming dual for a robust quadratic optimization problem with norm-constrained uncertain sets. Following the robust optimization methodology, we first introduce the robust counterpart of this robust quadratic optimization problem. Then, the authors obtain a second-order conic programming dual problem for this robust quadratic optimization problem. Moreover, by using a characteristic cone constraint qualification, the authors present a zero duality gap result between them.

A fuzzy adaptive control method based on command filtering technology is proposed for a stochastic system of flexible joint manipulators with dead-zone input, which achieves tracking control of the system output on the expected trajectory. Firstly, command filtering technology is used to solve the problem of "explosion of complexity" inherent in the traditional backstepping method, and error compensation mechanism is introduced to eliminate the influence of filtering errors on the system control precision. Then, a fuzzy logic system is utilized to deal with uncertainties and stochastic disturbances in the system, which overcomes the influence of stochastic disturbances and improves the control effect of the system. Finally, considering the system with dead-zone input, the control signal is constructed by backstepping control method, which conquers the adverse impact of dead-zone input on system performance. In the stability analysis, the effectiveness of the control strategy studied in the stochastic system of flexible joint manipulators with dead-zone input is proved, and it is verified by Matlab simulation.

In this paper, we study composite quantile regression(CQR) and variable selection of linear errors-in-variables models where the response and multi-dimensional covariates are mixed random missing. In order to improve the estimation efficiency, we propose the CQR estimator of regression coefficients based on inverse probability weighting and measurement error correction factor. The proposed CQR estimator can not only eliminate the influence of measurement errors on estimation results, but also deal with mixed random missing data effectively. At the same time, the asymptotic normality of the proposed estimator is obtained. Furthermore, a variable selection method based on the adaptive LASSO penalty is investigated for the measurement error models with mixed random missing data. The oracle property of the proposed penalized estimator is also established. Meanwhile, Monte Carlo simulation studies and a real data analysis are conducted to demonstrate the finite sample performance of the proposed methods.

The three-parameter Weibull distribution is one of the commonly used distributions for reliability analysis. However, its non-regular issue poses challenges to the validity of large sample properties of frequentist method. Additionally, the selection of prior in Bayesian estimation also faces certain issues. In order to provide practitioners with an alternative choice, this paper applies the generalized fiducial inference to the study of the three-parameter Weibull distribution. For the interest parameters such as reliability, generalized fiducial point estimation and confidence interval are constructed and compared with frequentist method and Bayesian method. Simulation results show that generalized fiducial point estimation has smaller or comparable mean square error and shorter or comparable average interval length while maintaining coverage probability. Finally, the effectiveness of generalized fiducial inference in the three-parameter Weibull distribution is demonstrated using data on single carbon fibers strength and ball bearings.

Digital transformation has gradually become an important driving force for enterprise development under the background of digital economy. Combining the problems of “weak digital technology foundation” and “difficult business empowerment of digital technology” from digital transformation, this paper divides digital transformation into digital technology resources and digital technology empowerment from the perspective of resources and capabilities to explore the antecedent (managerial resilience) that promote the solution of transformation problem and identify the economic consequence of transformation (firm performance). Based on the data of 557 firms listed in the Stock Exchange of Hong Kong, China from 2010 to 2020, the text analysis method is used to demonstrate that: 1) Managerial resilience has positive effects on both digital technology resources and digital technology empowerment, and has a stronger effect on digital technology empowerment; Industry competition negatively moderates the effect of managerial resilience on digital technology empowerment, but doesn't moderate the effect of managerial resilience on digital technology resources. 2) Digital technology empowerment has positive impacts on firm performance, while digital technology resources have no significant impact on firm performance. 3) Both “the promotion effect of managerial resilience on digital technology resources and digital technology empowerment” and “the positive impact of digital technology empowerment on firm performance” are stronger in large firms and non-manufacturing firms. 4) Managers with strong resilience promote the construction of digital technology and improve digital technology empowerment ability of firms by optimizing the human capital structure of management to help digital transformation; Digital technology empowerment enables firms to gain performance by improving operational efficiency and operating cost control efficiency; Managers with strong resilience can help firms acquire digital technology empowerment capabilities to improve firm performance. The conclusion of this paper provides important enlightenment for firms to implement digital transformation and obtain the economic benefits of digital transformation.

The price mechanism failure caused by fake quality information disclosure restricts the development of the agricultural product market. The transparent and traceable feature of information in the blockchain provides a new solution to the problem of fake quality information disclosure in the agricultural product supply chain. However, the high cost of blockchain, consumer preferences in the market, and block traceability accuracy all affect the strategy of adopting blockchain. This paper uses the signaling game and sets price as the signal to explore blockchain adoption strategies and quality information disclosure strategies in the agricultural product supply chain. This paper finds that: 1) The agricultural supermarket will adopt blockchain technology only when the information increase is high; when farmers' planting cost is low, the agricultural supermarket will be more willing to adopt blockchain. 2) When blockchain technology is not adopted and the planting cost is low, low-quality farmers have the incentive to deliberately set high prices to confuse the market. 3) The adoption of blockchain technology by the agricultural supermarket may harm farmers' profits. Only when the consumers' preference is more information-sensitive, all participants in the agricultural product supply chain can benefit from blockchain.

Aiming at the leader-follower intelligent networked vehicle cooperative formation coherence control problem, an observer-based distributed event-triggered control algorithm is proposed. Firstly, by analysing the vehicle dynamics, the vehicle state-space equations are modelled, based on which the system model is constructed and the trigger threshold is designed. Secondly, an observer is designed to solve the problems of system partial state unmeasurability, unknown perturbation and nonlinearity, and a distributed event-triggered control algorithm is proposed by utilizing the estimated states. The algorithm reduces the update frequency of the controller control signal by determining whether the trigger condition is satisfied, thus saving communication and computation resources. The proposed algorithm realizes the estimation of vehicle states, formation control, significantly improves the stability and reliability of the system, and enhances the cooperative efficiency of vehicle formation. Finally, the feasibility and effectiveness of the proposed method are described by simulation experiments on a four-vehicle leader-follower vehicle formation.

This paper introduces a noncooperative-cooperative biform game model under simultaneous model, electricity generator-leader model, and electricity retailer-leader model. It delves into the decision-making problem of new energy investment in the power system. The results show that: 1) The new energy investment under different power structures depends on the cost coefficient of traditional energy investment. 2) Both generators and retailers prefer to be leaders in the supply chains, aiming to maximize their individual profits. However, the overall profit of the supply chain is optimized in the simultaneous model. 3) An increase in the cost coefficient of new energy investment will reduce the investment in new energy and the profitability of generators and retailers, but the new energy investment cost-sharing ratio of the electricity generator will increase. 4) An increase in the preference coefficient of new energy will increase the investment in new energy and the new energy investment cost-sharing ratio of the electricity retailer will increase, thus increasing the profitability of both the generator and the retailer.

In this paper, the edge controllability of multi-agent systems under matrix weights is studied by using the transformation of topological graphs from point graphs to line graphs, where dynamics occur on the edges. Firstly, from the perspective of graph theory, a quantitative analysis is conducted on the incidence matrix of a line graph, and the relationship between the rank of the incidence matrix of the line graph and the number of connected components of the line graph are given. Furthermore, we find that there is a certain relationship between the algebraic multiplicity of zero eigenvalues of the Laplacian matrix of the line graph and the number of connected components of the line graph. Secondly, under the leader follower structure model, two conditions that need to be satisfied when the multi-agent system is edge controllable are obtained. In addition, according to the definition of canonical transformation, the balanced symbolic line graph of matrix weight structure is transformed into an unsigned line graph without negative edges. The results show that the controllability of the line graph before and after transformation is equivalent. Finally, the relationship between the controllability of line graphs and point graphs is analyzed, and it is found that when the point graph is structurally imbalanced, the controllability of line graphs is equivalent to that of point graphs.

Currently, one of the main challenges in practical modeling lies in the fact that training and testing data come from different distributions. Stable learning addresses this issue by decorrelating all covariates through sample reweighting, thereby achieving stable predictive performance. While machine learning methods such as stable learning show good results in experiments, there are still theoretical gaps, such as the lack of metrics for model stability under testing data and explanations for why stable learning maintains stable predictions across multiple environments. This paper proposes a new metric of stability, compares stable learning methods with ordinary least squares and explores the reasons why stable learning maintains stability across multiple environments. Finally, the paper validates the theory through simulated experiments. This research contributes to refining the theory of stability in stable learning, enhancing the understanding of stability in stable learning, and guiding the selection of practical modeling methods.