Linking online identities of users among countless heterogeneous network services on the Internet can provide an explicit digital representation of users, which can benefit both research and industry. In recent years, user identity linkage (UIL) through the Internet has become an emerging task with great potential and many challenges. Existing works mainly focus on online social networks that consider inconsistent profiles, content, and networks as features or use sparse location-based data sets to link the online behaviors of a real person. To extend the UIL problem to a general scenario, we try to link the web-browsing behaviors of users, which can help to distinguish specific users from others, such as children or malicious users. More specifically, we propose a Siamese neural network (NN) architecture-based UIL (SAUIL) model that learns and compares the highest-level feature representation of input web-browsing behaviors with deep NNs. Although the number of matching and nonmatching pairs for the UIL problem is highly imbalanced, previous studies have not considered imbalanced UIL data sets. Therefore, we further address the imbalanced learning issue by proposing cost-sensitive SAUIL (C-SAUIL) model, which assumes higher costs for misclassifying the minority class. In the experiments, the proposed model is robust and exhibits a good performance on very large, real-world data sets collected from different regions with distinct characteristics.

Vehicle evaluation parameters, which are increasingly of concern for governments and consumers, quantify performance indicators, such as vehicle performance, emissions, and driving experience to help guide consumers in purchasing cars. While past approaches for driving cycle prediction have been proven effective and used in many countries, these algorithms are difficult to use in China with its complex traffic environment and increasingly high frequency of traffic jams. Meanwhile, we found that the vehicle dataset used by the driving cycle prediction problem is usually unbalanced in real cases, which means that there are more medium and high speed samples and very few samples at low and ultra-high speeds. If the ordinary clustering algorithm is directly applied to the unbalanced data, it will have a huge impact on the performance to build driving cycle maps, and the parameters of the map will deviate considerable from actual ones. In order to address these issues, this paper propose a novel driving cycle map algorithm framework based on an ensemble learning method named multi-clustering algorithm, to improve the performance of traditional clustering algorithms on unbalanced data sets. It is noteworthy that our model framework can be easily extended to other complicated structure areas due to its flexible modular design and parameter configuration. Finally, we tested our method based on actual traffic data generated in Fujian Province in China. The results prove the multi-clustering algorithm has excellent performance on our dataset.

In recent years, new media have exacerbated the complexity of online public opinion scenarios through fragmentation of information, diversification of public opinion, rapid diffusion of public opinion, and concealment of information sources, which have posed several serious challenges to the benign development of online public opinion ecosystems. Therefore, based on diversified public opinion scenarios, we study the interaction between information dissemination and the evolution of group opinions and the dissemination laws to solve the problem of imprecise grasping of the dissemination laws in complex public opinion scenarios. Facing the two-way interaction between online platforms and real society, we constructed a coupled online–offline viewpoint evolution dynamics model, which considers factors such as the user subject level and the network environment level, and combines viewpoint dynamics theory with information dissemination dynamics theory. Based on the real case of dual interaction between online and offline, we carry out the construction of a two-layer coupling network and numerical fitting comparison experiments to study the synergistic and penetration mechanism of public opinion in both online and offline multi-spaces. Based on parametric analysis experiments, the influence of different factors on communication indicators is mined, and the driving effect of the viewpoint environment of offline communication on online public opinion is studied, which reveals the objective role of multi-factors on the law of intralayer communication, cross-network communication, and viewpoint evolution, and provides strategic suggestions for the comprehensive management of public opinion in online–offline large-scale mass incidents.

Each city can be divided into different functional areas for different purposes, such as commercial areas, residential areas, leisure areas, office areas, etc. How to correctly divide these areas is very helpful for urban self-cognition and urban planning. Mobile phones have become almost a must-have for everyday life nowadays in big cities. People carry their mobile phones all the time, making phone calls and surfing on the Internet. For this reason, the Open Information Dynamic Data (OIDD) is a more representative indicator of the mobility characteristics of the population. In this study, we use the user, time and location information extracted from OIDD data to analyze and divide the urban functional areas for the purpose of understanding the regional composition of the city. We use the Latent Dirichlet Allocation (LDA) model and incorporate time and space information into the Dirichlet distribution to participate in the model hypothesis. This allows for the model's high-level analysis capabilities to exploit potential urban area functions through human mobility patterns.

With the popularity of Internet services and the explosive growth of information, recommendation algorithm have become increasingly crucial in the all-media era. However, existing single-domain recommendation algorithms mainly rely on user-item interactions to construct user representations, which limits their ability to capture higher-order information about users and items. To address these challenges, we propose an entity-relation pair attention-based representation with knowledge graph for media-content cross-domain recommendation (EPAR-MCDR). In this framework, we adopt knowledge graph embedding to learning entity representation, which maps nodes and edges to a low-dimensional vector space to form an initial embedding of target domain. To overcome the limitations of single-domain data sparsity, we introduce a domain adaptation technique. It effectively utilizes the rich data in source domain through min–max game between the predictor and the user and item domain discriminator. The proposed method is compared and analyzed with three real-world datasets. Extensive experiments show that the method has effective improvement in the cross-domain recommendation.