BL-MNE: Emerging Heterogeneous Social Network Embedding Through Broad Learning with Aligned Autoencoder

Abstract

Network embedding aims at projecting the network data into a low-dimensional feature space, where the nodes are represented as a unique feature vector and network structure can be effectively preserved. In recent years, more and more online application service sites can be represented as massive and complex networks, which are extremely challenging for traditional machine learning algorithms to deal with. Effective embedding of the complex network data into lowdimension feature representation can both save data storage space and enable traditional machine learning algorithms applicable to handle the network data. Network embedding performance will degrade greatly if the networks are of a sparse structure, like the emerging networks with few connections. In this paper, we propose to learn the embedding representation for a target emerging network based on the broad learning setting, where the emerging network is aligned with other external mature networks at the same time. To solve the problem, a new embedding framework, namely "Deep alIgned autoencoder based eMbEdding" (DIME), is introduced in this paper. DIME handles the diverse link and attribute in a unified analytic based on broad learning, and introduces the multiple aligned attributed heterogeneous social network concept to model the network structure. A set of meta paths are introduced in the paper, which define various kinds of connections among users via the heterogeneous link and attribute information. The closeness among users in the networks are defined as the meta proximity scores, which will be fed into DIME to learn the embedding vectors of users in the emerging network. Extensive experiments have been done on real-world aligned social networks, which have demonstrated the effectiveness of DIME in learning the emerging network embedding vectors.

Collective Link Prediction Oriented Network Embedding with Hierarchical Graph Attention

“…• DIME [30]: An embedding framework for multiple heterogeneous aligned network with aligned autoencoders to transfer information and improve the link prediction in emerging networks. • MNN [1]: A multi-neural-network framework for intranetwork link prediction over aligned networks.…”

“…e study of multiple aligned networks provides a direction to alleviate the data insu ciency problem. Some research works propose to transfer information across networks by anchor links to enhance the link prediction results within multiple networks mutually [1,30,31]. Besides, many existing works aim at anchor link formation prediction [11,12,21].…”

Deep Collaborative Embedding for information cascade prediction

“…where X (m) andX (m) are the original cascading context matrix and the re- are straightforwardly fed into DCE. To overcome this issue, inspired by the idea used in the existing works [20,26] which assign more penalty (corresponding to larger weight) to the loss incurred by non-zero elements than that incurred by zero elements, the L x can be redefined as…”

How much topological structure is preserved by graph embeddings?

“…Researchers have proposed various approaches such as matrix factorization [4,7,50,72] and deep neural networks [63,5,30,22,35,34,64,11,69]. The topics are also varied, including unsuper-vised graph embedding [4,50,51,57,41,39], supervised graph embedding [74,61,63,41,21,30], community preserving embedding [68,6,78,10], and embedding in graphs of various types, such as the bipartite graphs [19], the heterogeneous graphs [28,7,56,14,18,65,76], the multi-relational graphs [52,45], the signed graphs [9,67,65], the uncertain graphs [24], the incomplete graphs [73], the dynamic graphs [77,36,79,…”