Introduction to the Special Issue on Advanced Graph Mining on the Web: Theory, Algorithms, and Applications: Part 2

We are delighted to present this special issue on Advanced Graph Mining on the Web: Theory, Algorithms, and Applications. Graph mining plays an important role in data mining on the Web. It can take full advantage of the growing and easily accessible big data resources on the Web, such as rich semantic information in social media and complex associations between users in online social networks, which is crucial for the development of systems and applications such as event detection, social bot detection, and intelligent recommendation. However, extracting valuable and representative information from Web graph data is still a great challenge and requires research and development on advanced techniques. The purpose of this special issue is to provide a forum for researchers and practitioners to present their latest research findings and engineering experiences in the theoretical foundations, empirical studies, and novel applications of Graph Mining. This special issue consists of two parts. In Part 2, the guest editors selected 10 contributions that cover varying topics within this theme, ranging from reinforced and self-supervised GNN architecture search framework to the streaming growth algorithm of bipartite graphs.

Wang et al. in “Contrastive Graph Similarity Networks” proposed a method for graph similarity learning, namely, Contrastive Graph Similarity Network. They make use of the complementary information of two input graphs and capture pairwise relations in a contrastive learning framework to enhance graph similarity learning, while developing a dual contrastive learning module with a node-graph matching and a graph-graph matching mechanism to reduce time complexity.

Qiao et al. in “A Dual-Channel Semi-Supervised Learning Framework on Graphs via Knowledge Transfer and Meta-Learning” propose a dual-channel framework for semi-supervised learning on graphs via knowledge transfer between independent supervised and unsupervised embedding spaces. They proposed a knowledge transfer head to bridge the gap between the generalization and fitting capability of the two models.

Zheng et al. in “Heterogeneous Information Crossing on Graphs for Session-based Recommender Systems” propose a novel graph-based method called Heterogeneous Information Crossing on Graphs, aiming to model heterogeneous user behaviors and capture the relationships between them in practical scenarios. And they also proposed an enhanced version, which incorporates contrastive learning technique to enhance item representation ability.

Luo et al. in “Semantic Interaction Matching Network for Few-shot Knowledge Graph Completion” focused on the few-shot learning in knowledge graph completion techniques and proposed a novel few-shot learning solution, named Semantic Interaction Matching network, which applies the Transformer framework to enhance the entity representation with capturing semantic interaction between entity neighbors.

Yu et al. in “Learning Neighbor User Intention on User-Item Interaction Graphs for Better Sequential Recommendation” proposed a Neighbor user Intention based Sequential Recommender, which utilizes the intentions of high-order connected neighbor users as high-order collaborative signals. And two modules are designed to fully exploit sequence features and model high-order collaborative signals for Sequential Recommendation.

Wang et al. in “Deep Adaptive Graph Clustering via von Mises-Fisher Distributions” proposed Deep Adaptive Graph Clustering via von Mises-Fisher distributions to solve challenges in node embedding in graph clustering. And they also proposed to update the cluster centers in an attraction-repulsion manner to make the cluster centers more separable.

Zhu et al. in “Incorporating A Triple Graph Neural Network with Multiple Implicit Feedback for Social Recommendation” propose to enrich the recommendation model by mining multiple implicit feedback and constructing a triple GCN component, aiming to solve the cold-start problem and data sparsity. And they leverage explicit feedback, social relationships and multiple implicit feedback to enhance the triple GCN component.

Kumar et al. in “Community Enhanced Link Prediction in Dynamic Networks” proposed a community enhanced framework to predict missing links on dynamic social networks. They first predict missing links using parameterized influence regions of nodes and their contribution to community partitions, and then generate a unique feature set using local, global, and quasi-local similarity-based as well as community information-based features.

Liu et al. in “Behavior Net: A Fine-grained Behavior-aware Network for Dynamic Link Prediction” propose a novel fine-grained behavior-aware network for dynamic network link prediction. It adapts a transformer-based graph convolution network to capture the latent structural representations of nodes by adding edge behaviors as an additional attribute of edges.

Xiao et al. in “PIDKG: Propagating Interaction Influence on the Dynamic Knowledge Graph for Recommendation” propose an influence propagation-enhanced deep coevolutionary method for recommendation, which can capture not only the direct mutual influence between interacting users and items but also influence propagation from multiple interacting nodes to their high-order neighbors at the same time on the dynamic knowledge graph.

The guest editors believe the articles appearing in this issue represent the frontiers of current topics in the field of Graph Mining and hope these articles will stimulate further development in this area. The editors sincerely appreciate the authors and reviewers’ tremendous contributions to this special issue.

We hope you enjoy this special issue and take some inspiration from it for your own future research.