User interactions on e-commerce platforms are inherently diverse, involving behaviors such as clicking, favoriting, adding to cart, and purchasing. The transitions between these behaviors offer valuable insights into user-item interactions, serving as a key signal for understanding evolving preferences. Consequently, there is growing interest in leveraging multi-behavior data to better capture user intent. Recent studies have explored sequential modeling of multi-behavior data, many relying on transformer-based architectures with polynomial time complexity. While effective, these approaches often incur high computational costs, limiting their applicability in large-scale industrial systems with long user sequences. To address this challenge, we propose the Transition-Aware Graph Attention Network (TGA), a linear-complexity approach for modeling multi-behavior transitions. Unlike traditional transformers that treat all behavior pairs equally, TGA constructs a structured sparse graph by identifying informative transitions from three perspectives: (a) item-level transitions, (b) category-level transitions, and (c) neighbor-level transitions. Built upon the structured graph, TGA employs a transition-aware graph Attention mechanism that jointly models user-item interactions and behavior transition types, enabling more accurate capture of sequential patterns while maintaining computational efficiency. Experiments show that TGA outperforms all state-of-the-art models while significantly reducing computational cost. Notably, TGA has been deployed in a large-scale industrial production environment, where it leads to impressive improvements in key business metrics.

Local differential privacy (LDP) mechanisms are widely used to collect data generated by IoT sensor devices to protect sensitive information. However, it easily leads to low data utility and high data computing cost due to complex structure and high dimensionality of IoT data. To alleviate this problem, we propose a High-dimensional Data Publishing method using Random responses based on Markov network (HDPRM). This method efficiently conducts the collection and analysis of high-dimensional data under the IoT architecture and satisfies LDP. In particular, it uses the expectation maximization (EM) algorithm to reconstruct the joint distribution of high-dimensional data attributes. Specifically, to improve the effectiveness of data release, we calculate the correlation between attributes and construct corresponding Markov network on the server. Additionally, we cluster high-dimensional attributes through the junction tree algorithm, filter out the joint probability that meets the requirements, and then use this probability to generate synthetic data for publication from the sampled data. Extensive experiments are conducted to comprehensively evaluate the performance of the HDPRM on three real-world datasets. The results show that the method achieves higher data utility under LDP guarantee compared to state-of-the-art methods.

The escalating threats posed by deepfakes to society and cybersecurity have triggered public anxiety, and growing efforts have been devoted to this pivotal research on deepfake detection. The generalization capability of existing models encounters a serious challenge. A prevailing explanation is that models tend to overfit artifacts in fake samples, thereby neglecting the exploration of available real ones. Prior studies have indicated that real images exhibit intra-class clustering and inter-class uniformity in the latent feature space, termed as compactness. Since deepfakes disrupt this property, exploring the common compactness of real samples may boost the generalization of models. In light of this, this paper proposes a targeted C ompact R econstruction L earning ( CRL ) strategy. It applies an enhanced Multi-View Reconstruction Loss (for self-compactness) to reconstruct only real images and a new Real-Sample Compactness Loss (for other-compactness) to bolster ties across real samples. Besides, a novel Joint - G uided R easoning ( JointGR ) module is introduced, which richly fuses features from the encoder-decoder and reconstructed differences. It fully capitalizes on multi-source features from CRL while improving the representational ability of our model. Under the latest benchmark, extensive experiments show our model keeps the competitive performance on most challenging datasets, even achieving state-of-the-art results on some. The code will be open-sourced at https://github.com/Dongyu-Han/CRL .

The epidemic of infectious diseases has a significant impact on society, the economy, and people’s lives. Social media, with its high user participation and rapid information dissemination, plays a crucial role in shaping public opinion. Fine-grained sentiment analysis of public opinion on infectious diseases can provide valuable insights for improving the quality of public services. However, there are few relevant studies on Chinese data due to language complexity and low resources. Moreover, most of the existing approaches utilize the Graph Neural Network (GCN) method by syntactic dependency trees to construct graphs of text, which ignore the potential link relationships between aspects and words. Therefore, to address this limitation, in this article, we propose a new method based on GCN using aspect-specific heterogeneous graphs, named ASHGCN, which combines BiLSTM, heterogeneous graphs, GCN, the mask and the attention mechanism. We mine social media posts related to COVID-19 for aspect-based sentiment analysis task (ABSA) for ten aspect entity types in both Chinese and English data. The heterogeneous graph is designed with two node types (aspect nodes and non-aspect nodes) and four edge connection types, including various relationships between aspect entities, and between aspect entities and non-aspect entities. In addition, we release a Chinese dataset and an English dataset that include medical and named entities, along with corresponding sentiment labels. Experiments on our datasets, as well as two public datasets, demonstrate that our method greatly improves performance in the ABSA task. Ablation experiments and case studies further support the effectiveness of the proposed approach.

The proliferation of location-aware devices has led to a wide-ranging applicability of trajectory data in diverse real-world scenarios. Despite the prevalent use of the k-means algorithm and differential privacy techniques in mainstream research for the generalization and protection of privacy-sensitive data, limitations persist in effectively reducing noise errors and enhancing overall algorithmic efficiency. This paper addresses the deficiencies observed in existing clustering methodologies applied to user trajectory data, focusing on data privacy preservation and enhanced utility. To this end, we propose a novel incremental clustering framework based on personalized differential privacy. The framework employs dynamic time warping for similarity assessment and incorporates temporal-based cluster protection mechanisms to fulfill privacy requirements. Moreover, it augments the Geo Indistinguishability (GI) privacy protection mechanism to tailor personalized privacy budgets. Subsequently, learning vector quantization is employed for incremental clustering synthesis of trajectory data, completing trajectory publication. Through experimental validation, our proposed model demonstrates significant improvements, with the data usability metric HD increasing by a minimum of 22% compared to existing algorithms, the privacy protection metric AMI improving by at least 18%, and the algorithm’s efficiency enhancing by no less than 20%.

Graph clustering is a fundamental task in graph analysis, and recent advances in utilizing graph neural networks (GNNs) have shown impressive results. Despite the success of existing GNN-based graph clustering methods, they often overlook the quality of graph structure, which is inherent in real-world graphs due to their sparse and multifarious nature, leading to subpar performance. Graph structure learning allows refining the input graph by adding missing links and removing spurious connections. However, previous endeavors in graph structure learning have predominantly centered around supervised settings, and cannot be directly applied to our specific clustering tasks due to the absence of ground-truth labels. To bridge the gap, we propose a novel method called homophily-enhanced structure learning for graph clustering (HoLe). Our motivation stems from the observation that subtly enhancing the degree of homophily within the graph structure can significantly improve GNNs and clustering outcomes. To realize this objective, we develop two clustering-oriented structure learning modules, i.e., hierarchical correlation estimation and cluster-aware sparsification. The former module enables a more accurate estimation of pairwise node relationships by leveraging guidance from latent and clustering spaces, while the latter one generates a sparsified structure based on the similarity matrix and clustering assignments. Additionally, we devise a joint optimization approach alternating between training the homophily-enhanced structure learning and GNN-based clustering, thereby enforcing their reciprocal effects. Extensive experiments on seven benchmark datasets of various types and scales, across a range of clustering metrics, demonstrate the superiority of HoLe against state-of-the-art baselines.

The curse of dimensionality in high-dimensional data makes it difficult to capture the abnormality of data points in full data space. To deal with this problem, we propose an outlier detection model based on Variational Autoencoder and Genetic Algorithm for subspace outlier analysis of high-dimensional data (VAGA). The proposed VAGA model constructs a variational autoencoder (VAE) to preliminarily detect outliers. Then the genetic algorithm (GA) is used to search the abnormal subspace of the outliers obtained by the VAE layer to provide a basis for subspace outlier analysis. The subsequent clustering of the abnormal subspaces help filter out the false positives which are fed back to the VAE layer to adjust network weights. The comparative experiments performed on three public benchmark datasets show that the outlier detection results of the proposed VAGA model are highly interpretable and have better accuracy performance than the state-of-the-art outlier detection methods.

In this poster, we propose a novel, effective tuple-based anonymization technique for categorical data over the Internet. By utilizing a new structure, called Candidate Encoding Sequence with Frequency, and a set of new rules for generating such a sequence for each domain value of the categorical data, we can effectively solve the key limitation of the existing methods. Our experimental results demonstrate the superiority of our method against the existing method in terms of the strength of privacy protection.