Independent Asymmetric Embedding for Information Diffusion Prediction on Social Networks

Xie, Wenjin; Wang, Xiaomeng; Jia, Tao

doi:10.48550/arxiv.2105.08291

Cited by 2 publications

(2 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[30] models the information propagation as heat diffusion in a high dimensional latent space through which the node's representation is learned. Following this work, [31,32] make modifications to improve the performance. The introduction of the deep learning method makes the use of network topology, temporal order and other features much more convenient, giving rise to a quick shift in designing the model.…”

Section: Related Workmentioning

confidence: 99%

CasSeqGCN: Combining Network Structure and Temporal Sequence to Predict Information Cascades

Wang¹,

Michalski²,

Ran³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

One important task in the study of information cascade is to predict the future recipients of a message given its past spreading trajectory. While the network structure serves as the backbone of the spreading, an accurate prediction can hardly be made without the knowledge of the dynamics on the network. The temporal information in the spreading sequence captures many hidden features, but predictions based on sequence alone have their limitations. Recent efforts start to explore the possibility of combining both the network structure and the temporal feature for a more accurate prediction. Nevertheless, it is still a challenge to efficiently and optimally associate these two interdependent factors. Here, we propose a new end-to-end prediction method CasSeqGCN in which the structure and temporal feature are simultaneously taken into account. A cascade is divided into multiple snapshots which record the network topology and the state of nodes. The graph convolutional network (GCN) is used to learn the representation of a snapshot. The dynamic routing and the long short-term memory (LSTM) model are used to aggregate node representation and extract temporal information. CasSeqGCN predicts the future cascade size more accurately compared with other state-of-art baseline methods. The ablation study demonstrates that the improvement mainly comes from the design of the input and the GCN layer. Taken together, our method confirms the benefit of combining the structural and temporal features in cascade prediction, which not only brings new insights but can also serve as a useful baseline method for future studies.

show abstract

Section: Related Workmentioning

confidence: 99%

CasSeqGCN: Combining Network Structure and Temporal Sequence to Predict Information Cascades

Wang¹,

Michalski²,

Ran³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Network embedding plays a crucial role in mining network data. It aims to represent the proximity between nodes by low-dimensional vectors, which can be achieved by different approaches such as adjacency matrix factorization [3,41], inferring the spreading sequence [2,36], learning the evolution of node status [10,18]. and more [6].…”

Section: Introductionmentioning

confidence: 99%

CoarSAS2hvec: Heterogeneous Information Network Embedding with Balanced Network Sampling

Zhan,

Jia

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Heterogeneous information network (HIN) embedding aims to find the representations of nodes that preserve the proximity between entities of different nature. A family of approaches that are wildly adopted applies random walk to generate a sequence of heterogeneous context, from which the embedding is learned. However, due to the multipartite graph structure of HIN, hub nodes tend to be overrepresented in the sampled sequence, giving rise to imbalanced samples of the network. Here we propose a new embedding method CoarSAS2hvec. The self-avoid short sequence sampling with the HIN coarsening procedure (CoarSAS) are utilized to better collect the rich information in HIN. An optimized loss function is used to improve the performance of the HIN structure embedding. CoarSAS2hvec outperforms nine other methods in two different tasks on four real-world data sets. The ablation study confirms that the samples collected by CoarSAS contains richer information of the network compared with those by other methods, which is characterized by a higher information entropy. Hence, traditional loss function applied to samples by CoarSAS can also yield improved results. Our work addresses a limitation of the random-walk-based HIN embedding that has not been emphasized before, which can shed light on a range of problems in HIN analyses.

show abstract

Independent Asymmetric Embedding for Information Diffusion Prediction on Social Networks

Cited by 2 publications

References 22 publications

CasSeqGCN: Combining Network Structure and Temporal Sequence to Predict Information Cascades

CasSeqGCN: Combining Network Structure and Temporal Sequence to Predict Information Cascades

CoarSAS2hvec: Heterogeneous Information Network Embedding with Balanced Network Sampling

Contact Info

Product

Resources

About