Information diffusion prediction is a fundamental yet challenging task in social networks. Existing approaches typically focus on either microscopic or macroscopic prediction, but few effectively integrate both perspectives. Moreover, current models often overlook the dynamic nature of cascades and struggle to capture both diffusion patterns simultaneously. This article addresses this limitation by proposing a novel multiscale information diffusion prediction model that unifies microscopic and macroscopic prediction based on graph attention networks (GATs) and hypergraph neural networks (MS-HGNN). Specifically, MS-HGNN integrates user social features from GATs and global cascade features from HGNN to predict the next affected user. A gated recurrent unit then generates a sequence of predicted users until reaching a virtual terminal user, enabling cascade size estimation. To enhance macroscopic prediction accuracy, we embed the model within a reinforcement learning framework and optimize it using policy gradient methods. Experimental results on four real-world datasets demonstrate that MS-HGNN consistently outperforms state-of-the-art baselines. On average, it achieves approximately 3%-4% improvements in Hits@k and 10%-30% improvements in MAP@k for microscopic prediction across datasets, while also reducing cascade size estimation error in macroscopic prediction.

Accurate prediction of public opinion trends during public health emergencies is crucial for understanding public attitudes and enabling proactive responses. Existing methods frequently exhibit inadequate prediction accuracy and elevated computational complexity in long-term forecasting. The proposed model is an enhanced time series transformer model that incorporates three key innovations. First, a sparse probabilistic attention mechanism reducing spatial complexity from O(L-2). Second, a progressive sequence decomposition architecture that explicitly separates trend and seasonal components. Third, a global attention distillation technique to mitigate error accumulation in autoregressive prediction. Experiments on a COVID-19 Weibo dataset containing over 780 000 posts demonstrate that the model accurately predicts trends up to seven times the input sequence length. The model outperforms existing methods by over 20% in terms of mean squared error (MSE) and mean absolute error (MAE). For a prediction length of 720, the model achieves an MSE of 0.457 and an MAE of 0.373, effectively capturing key fluctuation patterns and peak timings. The findings establish a substantial technical basis for public health management early-warning systems.

With the rapid development of the internet, the model of app updating and maintenance is shifting from enterprise-driven approaches to collaborative mechanisms powered by real-time user feedback. However, developers face challenges in handling massive, heterogeneous user reviews with dynamic demands. This article aims to address the challenge that existing research lacks a multidimensional evaluation system for prioritizing user reviews-which leads to developers' difficulty in allocating resources appropriately and responding to critical user needs promptly-and further seeks to provide a quantifiable decision-support tool for targeted app version optimization. This article proposes user review-based prioritization method (URPM), a novel approach that employs multidimensional evaluation to prioritize user reviews for app version optimization. The URPM framework operates in three stages: 1) quantifying user attention by clustering reviews and calculating topic proportions using the BERTopic model; 2) computing review sentiment scores using the SnowNLP tool; and 3) ranking reviews via a priority scoring function that integrates user attention, sentiment, timeliness, and ratings. Experimental results on eight music apps show that URPM achieves strong alignment between its generated high-priority reviews and actual developer update logs, with an F(hybrid )score of 0.625-0.702, NDCG@10 of 0.76-0.86, and NDCG@20 of 0.72-0.86. Comparative experiments on the QQ Music, NetEase Cloud Music, and KuGou Music datasets further demonstrate that URPM outperforms the second-best baseline across all three datasets, improving the F-hybrid score by 43.0%, 45.3%, and 38.2%, respectively. In addition, cross-domain validation confirms the robustness of URPM, yielding F(hybrid )scores of 0.655-0.717, NDCG@10 of up to 0.86, and NDCG@20 of up to 0.84.

With the rapid advancement of the internet, predicting user demand trends in user-generated content (UGC) on social media platforms can help businesses better understand user preferences, guiding decision-making and reform efforts. This paper explores product innovation by introducing a UGC-based user demand prediction technique. Initially, the BERTopic model is used to extract product attributes from UGC. The KANO model is then applied to categorize various user demands. An Attention-BiLSTM model, which incorporates empirical mode decomposition (EMD) and other features, is employed to forecast fluctuations and development trends in user demand preferences. The model’s performance is validated using different prediction datasets. To assess its effectiveness, the proposed hybrid model is compared to several leading deep learning algorithms. The combination of the KANO model and Attention-BiLSTM facilitates a more comprehensive analysis of sentiment and demand changes. Additionally, the limitation of existing sentiment-based trend prediction methods – unable to address long-range dependency problems – is overcome. The paper demonstrates the effectiveness of the proposed model framework using UGC data from “Auto-home”, highlighting the model’s superiority in prediction. Compared to state-of-the-art methods, this research improves online review analysis from a temporal perspective. This approach offers valuable insights for analyzing users’ product demand and predicting emotional trends related to products.

The rapid development of Internet of Things technology and the continuous improvement of consumer demand have spawned the emergence of intelligent products based on Internet of Things technology. The existing research on the closed-loop supply chain of intelligent products under the background of the Internet of Things still lacks consideration of different subsidy objects and recyclers. Therefore, for different subsidy objects in the three-stage closed-loop supply chain of intelligent products, a pricing decision model involving manufacturers, recyclers, retailers and consumers under the minimum standard supervision of recycling technology is constructed, and different recyclers are distinguished according to the characteristics of intelligent products. By comparing the profits of each subject in different models, the optimal model of government subsidy is obtained. It is found that under the four modes, subsidizing recyclers without self-recovery technology makes the total profit of the supply chain reach the optimal solution; capital recyclers can obtain the largest number of old intelligent product recycling by using self-recovery technology; when subsidizing manufacturers, the total profit and product recovery rate of the supply chain are at the bottom. In addition, recycling subsidies, subsidy impact coefficients and technical standards will affect the best profits of enterprises. Through the analysis of the closed-loop supply chain of intelligent products, the research provides theoretical support and decision-making basis for the supply chain members to formulate relevant strategies and the government to take relevant measures to promote resource utilization and environmental protection. Distinguished different recyclers by the characteristics of IOT products. Constructed game model under the supervision of the minimum recycling technology standards. Discussed the closed-loop supply chain decisions of different subsidy objects.

In the face of a sudden major epidemic, people's panic may likely lead to the disruption of the public opinion ecosystem and the disorder of public opinion order. Therefore, clarifying the key main bodies and mechanisms in governing online public opinion is of crucial significance for effectively managing and guiding it. Firstly, based on the sentiment analysis of opinion leaders, an evolutionary game model involving the government, netizens, and opinion leaders was constructed. It analyzed the gaming relationships among relevant stakeholders in the process of online opinion dissemination. Then, a simulation experiment is carried out to analyze the evolution of each stakeholder's strategy choice, and the effectiveness of the simulation scenario is verified by NLP technology. The research results show that when dealing with online public opinion during a major epidemic, the government should choose an appropriate time to intervene and reduce the cost of interfering with public opinion. The change in punishment intensity by the government has a greater impact on opinion leaders than on netizens. Additionally, when the government guides opinion leaders, increasing the degree of reward for opinion leaders is more effective than increasing the intensity of punishment.

Users are accustomed to posting their opinions on social platforms, and these text data can be used to analyze users preferences and personality traits. To predict the personality type of users, a user personality prediction model based on sentiment analysis and theme preference is proposed. This work employs latent Dirichlet allocation (LDA) to extract different topics of user comments and long-short term memory (LSTM) for text sentiment analysis. Then, the Big Five personality theory is introduced and combined with the text sentiment analysis method. Based on the data of the social event “28 years of commutation life”, the experimental results show that the accuracy of the sentiment classification on the validation set is 92%. Combined with the Big Five personality theory, the largest proportion of users’ emotions is “anger”, accounting for 23%. There are more users with extroverted personalities and neurotic personalities. These conclusions provide information in regard to predicting and controlling the direction of public opinion.

In order to capture and integrate the structural features and temporal features contained in social graph and diffusion cascade more effectively, an information diffusion prediction model based on the Transformer and Relational Graph Convolutional Network (TRGCN) is proposed. Firstly, a dynamic heterogeneous graph composed of the social network graph and the diffusion cascade graph was constructed, and it was input into the Relational Graph Convolutional Network (RGCN) to extract the structural features of each node. Secondly, the time embedding of each node was reencoded using Bi-directional Long Short-Term Memory (Bi-LSTM). The time decay function was introduced to give different weights to nodes at different time positions, so as to obtain the temporal features of nodes. Finally, structural features and temporal features were input into Transformer and then merged. The spatiotemporal features are obtained for information diffusion prediction. The experimental results on three real datasets of X (formerly known as Twitter), Douban, and Memetracker show that compared with the optimal model in the comparison experiment, the TRGCN model has an average increase of 4.16% in Hits@100 metric and 13.26% in map@100 metric. The validity and rationality of the model are proved.

People have been increasingly sharing information via social media platforms such as Twitter and Facebook in recent years, making understanding and predicting the spread of information a hot research topic. Accurate prediction of cascades within social platforms can effectively track the information dissemination process and prevent the spread of harmful information. Previous work either did not fully exploit the potential social graph structure or chose to predict communication sequences as a diffusion graph, resulting in inconsistency with the actual diffusion situation. We propose a method that is structurally simple and effective, requiring only sequential sequences of users' spread information and social relationships among users. Our method is based on LSTM (Long Short-Term Memory) for temporal feature extraction of cascade sequences and combined with GCN (Graph Convolutional Networks) for extracting features containing node topology in social graphs for microscopic prediction of information diffusion cascades. Comparative experi-ments on two real social platform datasets demonstrate that our method has a better prediction effect.

The issues arising from the evolution of big data and network environment have rendered the relationship among experts to no longer be an independent multi-attribute group decision-making process. In response, a PROMETHEE multi-attribute decision-making method based on data mining under dynamic hybrid trust network is proposed. This method incorporates evaluation similarity based on degree centrality and expert trust values based on K-hop centrality in the social network to determine the experts' weights. Furthermore, because the expert weight is somewhat affected by the attribute weight, the public-level attribute weight is obtained through data crawling and TF-IDF technology. The comprehensive attribute weight is acquired through a combination of expert-level attribute weights. Additionally, establishing a minimum adjustment cost model enables experts to follow a dynamic consensus reaching process in a composed hybrid trust network. Finally, to express the evaluation information more accurately in a complex linguistic environment, the probabilistic linguistic PROMETHEE method is introduced and applied to the site selection evaluation of charging and swapping stations. This highlights the feasibility and effectiveness of the proposed method through the comparison of decision-making methods and parameter sensitivity analysis.

A two-sided matching method with probabilistic linguistic term sets is proposed based on the BWM（best-worst method）, the DEMATEL（decision making trial and evaluation laboratory）, and the TODIM, to solve the product service modules matching problem with probabilistic language information considering its interaction and the psychological behavior of decision makers. Firstly, the weight of modules is determined using the BWM and the DEMATEL with probabilistic linguistic term sets. Then, the total dominance degree of product service modules is calculated based on the TODIM to obtain the satisfied matrix. Furthermore, a multi-objective optimization model is constructed to maximize the utility of products and services, which is converted into a single-objective model using the linear weighting method.Finally, an example of the new energy vehicle is provided to prove the effectiveness and feasibility of the proposed method,providing a new direction for the integrated development of new energy vehicle products and services.

The increased volume of rumors and attention on related topics during the COVID-19 pandemic have had a significant negative social impact. To combat rumors, it is crucial to study the actors involved in their spread. In this study, we first introduce prospect theory and construct an evolutionary game model between network operators and government regulators. We investigate collusion between network operators and Internet rumormongers as well as the regulatory behavior of government agencies. Second, we use prospect value to replace traditional expected utility to construct a profit prospect matrix and apply the dynamic replicator equation to analyze the equilibrium stability of the model. The stability conditions of the game between the two parties are closely related to the government’s regulatory costs, and the strength of government punishment after collusion is detected. Finally, we propose relevant countermeasures against collusion problems during the network rumor control period.

针对异质信息下被执行人财产隐匿行为甄别评估问题,考虑到被执行人评价指标关联性和执行人员的心理行为,提出一种基于复杂网络和前景理论的异质多属性决策方法.首先,对被执行人进行定量与定性分析,建立被执行人多维评价指标体系,进而利用复杂网络确定评价指标的权重;然后,考虑到被执行人评价指标的异质性,评价指标信息采用精确数、语言变量、概率语言等异质信息表示,在此基础上考虑决策者心理行为,利用前景理论计算各个被执行人的前景值,确定重点查控对象;最后,通过具体案例验证所提出方法的有效性和可行性,切实推动解决"执行难"问题,并为解决具有异质信息的实际管理决策问题提供新思路和新方法.

Mainstream social media, such as Facebook, Twitter, and Weibo, provide enterprises an opportunity to innovate and develop. User-generated content on social media platforms can help determine the needs of the user and identify a target market, providing a basis for enterprise innovation. In this study, we propose a user-interactive innovation knowledge acquisition model. Accordingly, the comments data on a selected forum were first crawled using network crawler software. Subsequently, we pre-processed the data to obtain a semi-structured user corpus. We then used the Latent Dirichlet Allocation model to cluster topics and obtain the subject words that were hidden from each comment text. A user demand ontology was built based on the subject words, and with an expert's reference, the product function ontology was established. Through semantic similarity matching, we integrated two ontologies to obtain the user-interactive innovation knowledge acquisition model. Finally, the model was validated using the Volvo XC60 automobile as an example. The empirical results showed that the proposed model could assist enterprises by providing ideas for follow-up innovation and product development.

This paper studies the impact of the decrease in government subsidies on the selection of the cooperation model of vehicle manufacturers’ in the new energy vehicle supply chain, and uses the mathematical modeling method to establish MA (cooperation between vehicle manufacturers and battery suppliers with better battery life), MB (cooperation between vehicle manufacturers and battery suppliers with similar battery life) MAB (vehicle manufacturer cooperates with both) and N (does not cooperate with both) four cooperation models. The results show that vehicle manufacturers and battery suppliers A and B have cooperation motives, but whether vehicle manufacturers prefer to cooperate with battery supplier A or battery supplier B is related to the decrease in subsidies. In addition, it studies the impact of the decrease in subsidies on the sales price, market demand and supply chain profit of new energy vehicles.

This paper constructs a new energy vehicle and traditional vehicle supply chain model composed of a vehicle manufacturer and an automobile retailer, compares the pricing, demand and supply chain profit of the two types of vehicles under decentralized decision-making and centralized decision-making, and then designs a revenue sharing + ex-factory price negotiation contract mechanism to coordinate the supply chain under certain conditions. The results show that: when the contract parameters meet certain conditions, the combination contract can achieve supply chain coordination and improve the profit of each member. In addition, we further study the impact of the perfection of charging facilities on the pricing, demand and supply chain profit of the two types of vehicles.

The rapid development of network promotes the development of multimedia teaching resources, which contains a large number of multimedia teaching resources. In order to effectively use these multimedia teaching resources, the importance of retrieval tools is highlighted. This paper designs and implements a teaching resource search system, which is used to search new media teaching resources related to education.