Deepfake technologies leverage deep learning to generate highly realistic videos involving face swapping and expression transfer, often exceeding the threshold of human visual perception. This poses serious challenges to social governance and digital security, highlighting the urgent need for reliable forgery detection methods. Training detection models without using real forgeries is considered a promising strategy to improve generalization. These approaches simulate diverse forgery traces to generate synthetic training data. However, most existing methods rely on 2 D image manipulation and fail to capture 3D forgery characteristics such as geometric distortion, expression mismatch, and texture anomalies-leading to poor performance on reconstruction-based forgeries. To address this problem, we propose a Reconstruction-Blended Image (RBI) generation method based on 3D Morphable Models (3DMM). By perturbing facial shape and expression parameters, this approach produces training samples that better reflect 3D reconstruction artifacts. When combined with traditional Self-Blended Images (SBI), the hybrid training strategy enhances the model's ability to detect a wider range of forgeries. Experiments show that this method improves AUC by$\mathbf{1 0. 7 9} \boldsymbol{\%}$on challenging cases like Face2Face forgeries. In summary, our 3D face reconstruction-based generation strategy significantly enhances the generalization and robustness of forgery detection models, offering a practical solution to emerging deepfake threats.

Knowledge graphs (KGs) are typically used to enhance the performance of recommender systems by leveraging their characteristic to supplement the sparse user-item interaction data in the latter. This is because KGs feature abundant semantic information about entities (e.g. users or items in recommender systems) and inter-entity relationships. Deep learning has been seen gain wide use in boosting the performance of recommender systems. Thanks to focusing merely on nodes in a KG, or entities, such an approach, however, suffers from well capturing the semantics hidden in the node connectivity when learning node embeddings, leading to undesirable discovery of potential semantic associations between nodes. Against this background, we propose for recommender systems a multi-level cross-view contrastive learning mechanism that learns high-quality feature representations from unlabeled data. By incorporating item-item correlations into item embeddings, we developed an interitem semantic association graph (SAG), which was intended to model more comprehensive and finer-grained inter-item semantic associations. A series of experiments have been carried out and the findings verify that the proposed mechanism outperforms other baseline models on all metrics, confirming the effectiveness of coupling SAGs with contrastive learning in enhancing the performance of recommender systems.

Graph Neural Networks (GNNs), which possess the ability to capture the connectivity and topological features between nodes in a graph structure, have been widely used in the field of Knowledge Graph (KG)-enhanced recommendation. However, the advantages of GNNs are largely limited by the quality of the introduced KG. For example, (1) in real recommendation scenarios, often a small number of popular items interact with a large number of users, which leads to sparse supervisory signals; (2) approaches using GNNs mostly focus on coarse-grained relationship modeling under the global structure and are susceptible to the interference of neighboring noisy information in the KG. Inspired by self-supervised learning, we propose a recommendation algorithm based on cross-view contrastive learning and apply it to a KG-enhanced recommendation task. By promoting mutual supervision of node features in different semantic spaces under local and global structures, the algorithm learns discriminative node representations and reduces noise interference during iterative propagation. Meanwhile, we utilize multi-task learning to promote co-training and shared learning between the recommendation task and contrastive learning to improve the overall generalization ability and the quality of recommendations. To verify the effectiveness of the proposed work, we conduct experiments on two public datasets, MovieLens and Book-Crossing. The experimental results show that our recommendation algorithm outperforms the baseline model in most cases in terms of AUC, recall@k, and F1, and the recommendation performance is improved by 1.7% -10.4% x.

During the pandemic, there has been a significant increase in the number of depression diagnoses. Meanwhile, the popularity of social media and the Myers-Briggs Type Indicator (MBTI) has provided new perspectives for designing lightweight and effective early warning methods for depression. These methods utilize personality traits and emotional characteristics obtained from social media posts to analyze users’ mental states and come to relevant conclusions. We therefore designed the PEER-MIND model, which classifies users’ depression status using MBTI labels obtained from their social media posts and emotion features determined by text and images from the posts as well. This model achieves a 3.58% higher prediction accuracy compared to baseline models when using fine-grained emotion detection. Also, we developed the TimeSlider-HMM model, an interpretable tool for assessing depression risk. This model evaluates changes in depression probability across different MBTI dimensions before and after significant depressive trends appear, providing us with insights into the interaction between personality and emotional factors in depression detection.

Representing and reasoning uncertain causalities have diverse applications in fault diagnosis for industrial systems. Owing to the complicated dynamics and a multitude of uncertain factors in such systems, it is hard to implement efficient diagnostic inference when a process fault occurs. The cubic dynamic uncertain causal graph was proposed for graphically modeling and reasoning about the fault spreading behaviors in the form of causal dependencies across multivariate time series. However, in certain large-scale scenarios with multiconnected and time-varying causalities, the existing inference algorithm is incapable of dealing with the logical reasoning process in an efficient manner. We, therefore, explore the solutions to enhanced computational efficiency. Causality graph decomposition and simplification, and graphical transformation, are proposed to reduce model complexity and form a minimal causality graph. An algorithm, event-oriented early logical absorption, is also presented for logical reasoning. It is mainly intended to minimize the computational costs for compulsory absorption operations in the early stage of reasoning process. The effectiveness of the proposed algorithm is verified on the secondary loop model of nuclear power plant by utilizing the fault data derived from a nuclear power plant simulator. The results show the anticipated capability of efficient fault diagnosis of the proposed algorithm for large-scale dynamic systems.

With the development of the Web, social media, the SINA Microblog for instance, is gaining wider popularity. This indicates most users' attitudes towards emerging technologies that happened recently. To obtain the data on social media, the traditional Web crawler is only able to obtain part of the information by collecting data in a non-login status because the crawler is not provided with the capabilities to log into the system. To enable information sharing among users, the SINA Microblog Open Platform offers a number APIs in which certain limitations such as those on the number of user requests of the microblog server, still exist. In this paper, a method that simulates user login to collect Microblog data related to hot topics on the SINA platform is proposed. What's more, the method overcomes the limitations of the SINA Microblog APIs, thus leading to getting more data from the system. In addition, the typical user behaviors, through the collected data, is found when they interact with social media.

Summary The precise prediction of bus routes or the arrival time of buses for a traveler can enhance the quality of bus service. However, many social factors influence people's preferences for taking buses. These social factors may include heavy traffic cost, traffic congestion, poor air quality and so forth. Existing prediction techniques rarely consider social sensing when predicting the bus arrival time. Accordingly, this paper proposes a social sensing enhanced service for predicting bus routes, which integrates sensing ability and social networks to understand and measure the influence between social events and vehicle velocity. We focus on the analysis of two different attributions: PT service quality attributions PEAs and road condition attributions PRCAs. Both of them synthesize the social sensing in their evaluation of bus routes. PEA represents individual preferences and PRCA represents physical factors that significantly influence vehicle velocity. Bus relevant social events were further categorized into PEA events or PRCA events. PEAs of buses were scored according to the tendency of bus conditions reflected in social events. Furthermore, an artificial neural network prediction model is established to estimate the bus travel time. Copyright © 2015 John Wiley & Sons, Ltd.

A recent study came up with an ad-hoc Semantic Web data sharing system. The system was designed for users to share Semantic Web data in an ad-hoc network without any central sharing service. For this purpose, the study proposed the storage format, network structure, and lookup method for the system. In this paper, we demonstrate how to implement a simulation of the Semantic Web data sharing system based on NS3, which is a full-range open-source network simulator, to evaluate the performance at the network level, and to find out certain issues we ought to address in the future for the system.

Chord is a well-established and classical Peer-to-Peer (P2P) protocol for its simplicity and high search efficiency. There are many studies about how to further improve Chord search efficiency. In this paper, we first discuss the finger table to analyze the reason why Chord has high search efficiency. Based on the analysis, we made a modification on the item start in the finger table of Chord to improve search efficiency. Results from theoretical analysis and experiments show that the modification improves search efficiency as we anticipated.

The key element to support ad-hoc resource sharing on the Web is to discover resources of interest. The hypermedia paradigm provides a way of overlaying a set of resources with additional information in the form of links to help people find other resources. However, existing hypermedia approaches primarily develop mechanisms to enable resource sharing in a fairly static, centralized way. Recent developments in distributed computing, on the other hand, introduced peer-to-peer (P2P) computing that is notable for employing distributed resources to perform a critical function in a more dynamic and ad-hoc scenario. We investigate the feasibility and potential benefits of bringing together the P2P paradigm with the concept of hypermedia link services to implement ad-hoc resource sharing on the Web. This is accomplished by utilizing a web-based Distributed Dynamic Link Service (DDLS) as a testbed and addressing the issues arising from the design, implementation, and enhancement of the service. Our experimental result reveals the behavior and performance of the semantics-based resource discovery in DDLS and demonstrates that the proposed enhancing technique for DDLS, topology reorganization, is appropriate and efficient.

We describe the design of a flood warning system which uses a set of sensor nodes to collect readings of water level and a grid-based flood predictor model developed by environmental experts to make flood predictions based on the readings. The reporting frequency of sensor nodes is required to be adaptive to local conditions as well as the flood predictor model to optimize battery consumption. We therefore propose an energy aware routing protocol which allows sensor nodes to consume energy according to this need. This system is notable both for the adaptive sampling regime and the methodology adopted in the design of the adaptive behavior, which involved development of simulation tools and close collaboration with environmental experts