Clustering social audiences in business information networks

Zheng, Yu; Hu, Ruiqi; Fung, Sai-fu; Yu, Celina P.; Long, Guodong; Guo, Ting; Pan, Shirui

doi:10.1016/j.patcog.2019.107126

Cited by 22 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition to static charts, there are also tech-nologies applied to dynamic graphs [21]. In terms of applications, HIN embedding can be applied to many aspects, such as recommender systems [10], business networks [22], and bioinformatics networks [23].…”

Section: Related Workmentioning

confidence: 99%

HeteEdgeWalk: A Heterogeneous Edge Memory Random Walk for Heterogeneous Information Network Embedding

Liu

Zhang

et al. 2023

Entropy

View full text Add to dashboard Cite

Most Heterogeneous Information Network (HIN) embedding methods use meta-paths to guide random walks to sample from HIN and perform representation learning in order to overcome the bias of traditional random walks that are more biased towards high-order nodes. Their performance depends on the suitability of the generated meta-paths for the current HIN. The definition of meta-paths requires domain expertise, which makes the results overly dependent on the meta-paths. Moreover, it is difficult to represent the structure of complex HIN with a single meta-path. In a meta-path guided random walk, some of the heterogeneous structures (e.g., node type(s)) are not among the node types specified by the meta-path, making this heterogeneous information ignored. In this paper, HeteEdgeWalk, a solution method that does not involve meta-paths, is proposed. We design a dynamically adjusted bidirectional edge-sampling walk strategy. Specifically, edge sampling and the storage of recently selected edge types are used to better sample the network structure in a more balanced and comprehensive way. Finally, node classification and clustering experiments are performed on four real HINs with in-depth analysis. The results show a maximum performance improvement of 2% in node classification and at least 0.6% in clustering compared to baselines. This demonstrates the superiority of the method to effectively capture semantic information from HINs.

show abstract

Section: Related Workmentioning

confidence: 99%

HeteEdgeWalk: A Heterogeneous Edge Memory Random Walk for Heterogeneous Information Network Embedding

Liu

Zhang

et al. 2023

Entropy

View full text Add to dashboard Cite

show abstract

“…I N recent years, graphs have attracted a surge of research attention with the development of networked applications in social networks [1], human knowledge networks [2], business networks [3] and cybersecurity [4]. However, the bulk of the existing researches focus on static graphs [5], [6], yet the real-world graph data often evolves over time [7], [8].…”

Section: Introductionmentioning

confidence: 99%

Anomaly Detection in Dynamic Graphs via Transformer

Liu¹,

Pan²,

Wang³

et al. 2023

IEEE Trans. Knowl. Data Eng.

Self Cite

View full text Add to dashboard Cite

Detecting anomalies for dynamic graphs has drawn increasing attention due to their wide applications in social networks, e-commerce, and cybersecurity. Recent deep learning-based approaches have shown promising results over shallow methods. However, they fail to address two core challenges of anomaly detection in dynamic graphs: the lack of informative encoding for unattributed nodes and the difficulty of learning discriminate knowledge from coupled spatial-temporal dynamic graphs. To overcome these challenges, in this paper, we present a novel Transformer-based Anomaly Detection framework for DYnamic graphs (TADDY). Our framework constructs a comprehensive node encoding strategy to better represent each node's structural and temporal roles in an evolving graphs stream. Meanwhile, TADDY captures informative representation from dynamic graphs with coupled spatial-temporal patterns via a dynamic graph transformer model. The extensive experimental results demonstrate that our proposed TADDY framework outperforms the state-of-the-art methods by a large margin on six real-world datasets.

show abstract

“…In the unsupervised learning domain, clustering is a fundamental data analysis task which partitions a group of objects in such a way that, according to some principles, similar objects are assigned to a partition while different partitions contain dissimilar objects. Clustering has extensive real-world applications, e.g., detecting communities among the social media networks [1], segmenting images via clustering the semantic features [2], and helping enterprises discovering the social audience clusters in a business information network [3]. Clustering algorithms have continuously emerged in the past few decades.…”

Section: Introductionmentioning

confidence: 99%

“…Clustering methods that take advantage of the deep learning techniques for jointly learning hidden features of the data are referred to as the deep clustering (DC) models [16]. Typically, DC algorithms can be classified into three categories: (1) direct cluster optimization; (2) autoencoder based models; (3) generative model based methods. The direct cluster optimization DC models only focus on the clustering loss to optimize the deep neural network, such as [17][18][19] leverage convolutional neural networks to uncover the representation and exploit affinities between data points for guiding the clustering procedure.…”

Section: Introductionmentioning

confidence: 99%