Data pricing is a key link to promote the efficient circulation of data in the market. However, the existing methods are still insufficient in terms of pertinence, dynamism and comprehensiveness. Therefore, we proposed a data pricing prediction model based on sparrow search optimization XGBoost, aiming to provide a reference for pricing decisions in data market. First, we crawled the data transaction information of Youedata.com and performed preprocessing operations such as outlier processing, one hot encoding and logarithmic transformation on the dataset; Secondly, we conducted exploratory data analysis to understand the distribution of data and their correlation. Then, we used the LASSO algorithm to select features for the dataset and constructed a data pricing prediction model based on SSA-XGBoost. Finally, we compared and analyzed it with six machine learning models including LightGBM, GBDT, MLP, KNN, LR and XGBoost. The experimental results show that in terms of the R-squared, the prediction results of the proposed SSA-XGBoost model exceed the above six models by 4.9%, 7.4%, 7.1%, 23.8%, 12.8%, and 2.3% respectively, and are superior to the state-of-the-art work. Furthermore, the evaluation results of the five indicators of MSE, RMSE, MAE, MAPE, and RMSPE are better than other models, showing higher stability.

Since the COVID-19 pandemic began in late 2019, economic uncertainty and supply chain disruptions have significantly affected China's real economy. This has reduced consumer reliance on physical stores, increasing market and operational risks for brick-and-mortar retailers. This study analyzes the impact of the pandemic on shopping center performance using operational data from 7, 010 stores in China alongside COVID-19-related data. The analysis focuses on two indicators: Rent levels and lease termination rates, considering both rental and tenant perspectives. Findings reveal that COVID-19 negatively impacts retail rents and lease termination rates, with regions experiencing severe outbreaks facing greater declines. During lockdowns, the negative effects on rents and termination rates were more pronounced than during the broader pandemic period. Additionally, stores with sales-based rent experience a greater adverse impact compared to those without. The study concludes that the pandemic significantly reduces store rent while the likelihood of early lease terminations remains relatively low. These empirical results offer valuable insights for shopping center operational management and crisis response strategies.

To address the issue of revenue distribution between government departments and enterprises in the operation of public data authorization, an evolutionary game model was constructed for both government and enterprise parties. The impacts of different incentive levels and revenue distribution ratios on the strategic choices and evolutionary trends of both government and enterprise were analyzed. It was found that when the government chose a strategy of weak authorization and strong regulation, enterprises showed a higher tendency to actively participate in public data sharing. In addition, when the revenue distribution ratio between government and enterprise was 3:7, the game evolution of both parties tended to be stable, reaching a balanced state that is beneficial and sustainable for both parties.

With the rapid advancement of information technology, the sharing of educational resources has become an integral component of the modern educational system. However, traditional data elements in educational resources face challenges such as data security, copyright protection, and trust mechanisms. Blockchain technology, as a distributed ledger technology, offers innovative solutions for the sharing of educational resources data elements with its characteristics of immutability, decentralization, and transparency. This paper designs an educational resource sharing platform based on blockchain technology, and experiments have demonstrated that by optimizing the blockchain threshold elimination model, the overall performance of the platform can be enhanced, improving the security of data processing and achieving satisfactory results.

In the domain of data governance, crimes involving virtual currencies have emerged as an integral concern that cannot be overlooked. To address challenges such as the difficulty of evidence collection and the low probability of recovering stolen funds in virtual currency crimes, this paper proposes a new mechanism for the electronic storage and retrieval of evidence using blockchain technology, elaborating on its core steps and underlying technology. Moreover, from the perspective of virtual currency transaction intermediaries, this study employs game theory to analyze the issue, constructing replicator dynamics equations, solving for the Jacobian matrix, and exploring the direction of game evolution and the factors influencing the decision-making of the participants. This analysis demonstrates that the decision-making of virtual currency criminals is impacted by this electronic evidence mechanism, which can deter illicit intermediaries from assisting in money laundering activities, thereby reducing the feasibility of committing crimes with virtual currencies. Lastly, the paper offers policy recommendations to enhance the implement ability of the evidence storage and retrieval mechanism in regulating virtual currency crimes.

As the era of large-scale highway maintenance arrives, the maintenance strategies have transitioned to a holistic approach that prioritizes safety, economic feasibility, and environmental sustainability. This research introduces a multi-objective optimization model for highway maintenance that incorporates the interplay of decision-maker preferences across three key objectives: Highway safety performance, maintenance engineering cost, and carbon emissions. This study employs a large-sample data analysis on a subset of the Lianhuo Highway network, which includes 2,842 pavement sections. This approach mitigates the impact of outliers, ensuring a substantial data buffer that fortifies the model's capacity for generalization and bolsters its robustness. The findings reveal a Pareto-optimal relationship among the three scrutinized variables. A particularly noteworthy observation is the M-shaped trajectory of carbon emissions, which initially rise, then decline, and ultimately rebound, contingent upon the selected maintenance strategy. Furthermore, an examination of the relationship between maintenance costs and safety performance discloses a trend of diminishing marginal returns, illustrating that the incremental gains in safety performance attenuate as maintenance investment escalates.

In July 2024, a shooting incident involving President Trump drew widespread public attention, highlighting the need for a deeper understanding of political rhetoric among U.S. Congressional members. This study analyzes discourse patterns on Twitter using large language models (LLMs), specifically ChatGPT and bidirectional encoder representations from transformers (BERT), to explore underlying factors that may contribute to polarizing political language. By collecting and preprocessing Twitter data, we initially labeled 20, 000 tweets using ChatGPT and then utilized the BERT-large model to classify the remaining 980, 000 tweets. The analysis identified party affiliation and geographic region as significant factors influencing political rhetoric. Republican lawmakers exhibited a higher prevalence of polarizing language, while New Jersey recorded the highest rate among the states. Newly elected Congressional members also tended to adopt more provocative language, potentially as a strategy to engage with their voter base or distinguish themselves in a competitive political environment. Temporal analysis revealed spikes in polarizing rhetoric corresponding to events such as discussions on the new fiscal year budget. This study offers insights into the dynamics of political discourse, providing a foundation for promoting constructive dialogue and fostering institutional resilience.

This article investigates the structural transformation of Ukraine's financial market within the context of innovative metaspace technologies. It presents models outlining the structural changes in financial market sectors amid the rapid development of metaspace technologies, emphasizing the systematic prognostic nonlinear dependence of deposit, credit, and currency flows on time intervals within the FinTech services' payment platform. The article introduces system methods for modeling nonlinear processes and combinatorial procedures for organizing the environment of innovative metaspace technologies within the financial market. It systematically categorizes approaches for measuring the effectiveness signs of financial market structure development in the innovative metaspace. The article argues for the need to structure and evaluate the financial market sectors of Ukraine through non-combinatorial tasks, emphasizing the convenience of describing variables influencing market structure and visually depicting the metaspace environment as a simulated FinTech ecosystem. Standardizing performance indicator measurement algorithms for precise intervals of financial resource movement is advocated. The article establishes that combinatorial tasks for structuring financial market development objects in the metaspace environment should align with configurations influenced by variable factors, such as currency, deposit, and credit flows. An analysis of transactions involving currency exchange, deposits, and loans through FinTech services' payment platform in Ukraine's financial market is presented. Additionally, the article conducts a formalized assessment of changes in currency, deposit, and credit flow transactions influenced by the digital environment of FinTech services in Ukraine's financial market sectors.

Drug-protein interaction (DPI) prediction in drug discovery and new drug design plays a key role, but the traditional in vitro experiments would incur significant temporal and financial costs, cannot smoothly advance drug-protein interaction research, so many computer prediction models have emerged, and the current commonly used is based on deep learning method. In this paper, a deep learning model Computer-based Drug-Protein Interaction CBSG_DPI is proposed to predict drug-protein interactions. This model uses the protein features extracted by the Computed Tomography CT and Bert method and the drug features extracted by the SMILES2Vec method and input into the graph convolutional neural network (GCN) to complete the prediction of drug-protein interactions. The obtained results show that the proposed model can not only predict drug-protein interactions more accurately but also train hundreds of times faster than the traditional deep learning model by abandoning the traditional grid search algorithm to find the best parameters.

Given the rise of artificial intelligence, big data analytics has emerged as an important tool for processing and assimilating the enormous volume of data available on social media. It is of great theoretical and practical significance to explore the public opinion diffusion process and characteristics, and users' emotions of mega sports events based on big data statistics in the social media environment. This paper takes the Jakarta Asian Games, Russian World Cup and PyeongChang Winter Olympics held in 2018 as cases, uses text mining and social network analysis methods to analyze the dissemination process of social media users' data, presents the semantic words disseminated in sports events through high-frequency word cloud diagrams, and summarizes the general rules of public opinion dissemination. The results show that the more users' participation, the greater diffusion volume, and the diffusion process shows fast increasing, short duration, scattered topics, diversified contents, and strong guidance and weak continuity of attention. The high-frequency words, except for the names of the events, such as "cheer", "win the game" and "must win", have obvious concentration of emotional words.

With the rapid development of artificial intelligence technology and its advantages in decision-making, smart decision-making is receiving increasing attention from both academics and practitioners. This paper uses a bibliometric approach to analyze 783 works of literature related to smart decision-making from 1965-2022 to understand the research content and development process in the field. Through co-citation analysis, this paper identifies critical publications and research clusters in the field. Combined with keyword analysis, this paper provides a systematic review of the current state of research on smart decision-making. On this basis, the conceptual framework of the field is presented. According to the existing literature and bibliometric analysis, this study provides a list of research questions from three dimensions that deserve further consideration, providing insights into the trend of using AI for decision-making in the future.

Artificial intelligence has transformed marketing and consumer lives. However, despite its multiple beneficial effects, unprecedented artificial intelligence uses pose enormous challenges to consumer privacy protection. This research aims to provide a comprehensive picture of privacy research and enhance privacy protection. We first give more specific definitions of consumer privacy according to data analytic processes. Second, we investigate how AI-related factors, privacy risk, privacy protection, and other factors influence consumer privacy decision-making, and draw a conceptual model based on the extended Antecedents-Privacy Concerns-Outcomes (APCO) model. Third, we examine the potential risks of each data analytic step and the causes of privacy risks from consumer, company, and technology perspectives, and propose privacy protection schemes correspondingly. Lastly, we also provide agenda for future research.

This paper investigates the rank reversal issue in the selection of new energy vehicle types, focusing on consumers aged 20 to 30. It employs both the AHP and the PCbHA methods to rank four types of the new energy vehicles — pure electric vehicles, plug-in hybrid electric vehicles, range-extended electric vehicles, and fuel cell vehicles, based on ten influential factors: purchase cost, maintenance cost, fuel and electricity cost, safety, passability, endurance, appearance, brand power, power, and space. To verify the effectiveness of the PCbHA method in addressing the rank reversal problem, one alternative option is removed, and the ranking is recalculated with subsequent analysis of the results. The study finds that rank reversals often stem from the closeness of alternative weights. Through sensitivity analysis, this research reveals the impact of endurance attribute weight on decision outcomes, indicating that when the endurance weight reaches 0.35, the ranking of pure electric vehicles and range-extended electric vehicles reverses.

Connected and autonomous vehicles (CAVs) are expected to coexist alongside human-driven vehicles on roads for the foreseeable future. This study explores the stability and safety of mixed traffic streams, including traditional trucks and cars alongside CAVs. The study utilizes the intelligent driver model and cooperative adaptive cruise control model to characterize human-driven vehicles (including cars and trucks) and CAVs, respectively. It investigates how different ratios of trucks and penetration rates of CAVs impact the linear stability of mixed traffic flows and delineate their stability domains. Additionally, a simulation experiment is conducted using SUMO software to assess the safety implications of traffic congestion at on-ramp bottlenecks, specifically analyzing the safety dynamics of mixed traffic streams. The findings indicate that CAVs enhance both the stability and safety of mixed traffic flows. The presence of trucks is associated with reduced stability values at similar CAVs penetration rates. In scenarios without trucks, CAVs can elevate traffic safety by 58.28%-71.28%, whereas in the presence of trucks, although the enhancement diminishes, safety levels can still improve by 48.67%-65.11%.

This paper proposes a new discrete-time Geo/G/1 queueing model under the control of bi-level randomized (p, N₁, N₂)-policy. That is, the server is closed down immediately when the system is empty. If N₁ (≥1) customers are accumulated in the queue, the server is activated for service with probability p (0≤ p≤1) or still left off with probability (1-p). When the number of customers in the system becomes N₂ (≥ N₁), the server begins serving the waiting customers until the system becomes empty again. For the model, firstly, we obtain the transient solution of the queue size distribution and the explicit recursive formulas of the stationary queue length distribution by employing the total probability decomposition technique. Then, the expressions of its probability generating function of the steady-state queue size and the expected steady-state queue size are presented. Additionally, numerical examples are conducted to discuss the effect of the system parameters on some performance indices. Furthermore, the steady-state distribution of queue length at epochs n^-, n and outside observer's observation epoch are explored, respectively. Finally, we establish a cost function to investigate the cost optimization problem under the constraint of the average waiting time. And the presented model provides a less expected cost as compared to the traditional N-policy.