Friend Recommendation Based on Multi-Social Graph Convolutional Network

Chen, Liang; Xie, Yuanzhen; Zheng, Zibin; Zheng, Huayou; Xie, Jingdun

doi:10.1109/access.2020.2977407

Cited by 39 publications

(13 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As there exist three kinds of relations, we propose to fuse the multiple relations into a single relation with three fusion strategies. Inspired by [32], we take the symmetric normalized Laplacian matrices of the graph for the fusion of multiple relations, which can enrich the structural information propagated through nodes. Specifically, the symmetric normalized Laplacian matrix of the composition relation A C is defined as…”

Section: Graph Convolutionmentioning

confidence: 99%

Multi-Relational Graph Convolution Network for Service Recommendation in Mashup Development

Gao¹,

Wu²

2022

Applied Sciences

View full text Add to dashboard Cite

With the rapid development of service-oriented computing, an overwhelming number of web services have been published online. Developers can create mashups that combine one or multiple services to meet complex business requirements. To speed up the mashup development process, recommending suitable services for developers is a vital problem. In this paper, we address the data sparsity and cold-start problems faced in service recommendation, and propose a novel multi-relational graph convolutional network framework (MRGCN) for service recommendation. Specifically, we first construct a multi-relational mashup-service graph with three types of relations, namely composition relation, functional relation, and tagging relation. These three relations are indispensable and complement each other for capturing multi-view information. Then, the three relations in the graph are seamlessly fused with various strategies. Next, graph convolution is performed on the fused multi-relational graph to capture the high-order relational information between mashups and services. Finally, the relevance between mashup requirements and services is predicted based on the learned features on the graph. We conduct extensive experiments on the ProgrammableWeb dataset and demonstrate that our proposed method can outperform state-of-the-art methods in recommending services when only mashup requirements are available.

show abstract

Section: Graph Convolutionmentioning

confidence: 99%

Multi-Relational Graph Convolution Network for Service Recommendation in Mashup Development

Gao¹,

Wu²

2022

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…Ding et al (2017) also proposed a model called BayDNN based on structural features and used a Bayesian ranking to recommend new friends. Chen et al (2020) used a convolutional neural network (CNN) for user embedding in a graph convolutional network to process users and their neighbors' features. They recommended new friends based on the Bayesian ranking.…”

Section: Background and Related Workmentioning

confidence: 99%

SAFER: Social Capital-Based Friend Recommendation to Defend against Phishing Attacks

Guo

Cho

Chen

et al. 2022

ICWSM

View full text Add to dashboard Cite

The tremendous growth of social media has been accompanied by highly advanced online social network (OSN) technologies. Such advanced technologies have been heavily utilized by perpetrators as convenient tools for deceiving people in online worlds. Social capital has been discussed as a powerful mechanism to leverage interpersonal relationships in social networks in order for an individual to achieve his/her goal. The beauty of social capital is the ability to materialize non-monetary, less costly, and non-economic resources into tools to solve social problems. In this paper, we aim to leverage social capital (SC) to minimize online users' vulnerabilities to online deception. In particular, we propose a Social cApital-based FriEnd Recommendation scheme, called SAFER, that can protect OSN users from phishing attacks. We quantify three dimensions of social capital, namely, structural, cognitive, and relational, based on user features obtained from real datasets and model a user's friending behavior based on their social capital. In addition, to model a user's behavior upon being attacked by a phishing attacker, we developed the so-called SER-SEIR (Susceptible, Exposed, Recovered-Susceptible, Exposed, Infected, and Recovered) model as a variant of the SEIR model. Via extensive simulation experiments based on two real datasets considering bot-based and human-based attackers performing phishing attacks, we demonstrate the performance of four SC-based friend recommendation schemes with three non-SC-based comparable counterparts in terms of the ratio of detecting attackers and the fraction of users in the states of S, E, I, and R. Based on the performance comparison, we analyze the overall trends of their performance in terms of the extent of resistance against phishing attacks by bot or human attackers.

show abstract

“…Liao et al (2015) explored the friend recommendation issue in virtual worlds and developed an approach based on a hybrid SVM classifier considering user similarity and virtual contact strengths. Chen et al (2020) developed a framework that learned the potential features of higher-order neighbors based on multiple social networks and generated personalized friend recommendations using the improved graph convolutional network (GCN).…”

Section: Machine Learning-based User Recommendationmentioning

confidence: 99%

User recommendation in online health communities using adapted matrix factorization

Yang

Gao

2021

INTR

View full text Add to dashboard Cite

PurposeOnline health communities (OHCs) are platforms that help health consumers to communicate with each other and obtain social support for better healthcare outcomes. However, it is usually difficult for community members to efficiently find appropriate peers for social support exchange due to the tremendous volume of users and their generated content. Most of the existing user recommendation systems fail to effectively utilize the rich social information in social media, which can lead to unsatisfactory recommendation performance. The purpose of this study is to propose a novel user recommendation method for OHCs to fill this research gap.Design/methodology/approachThis study proposed a user recommendation method that utilized the adapted matrix factorization (MF) model. The implicit user behavior networks and the user influence relationship (UIR) network were constructed using the various social information found in OHCs, including user-generated content (UGC), user profiles and user interaction records. An experiment was conducted to evaluate the effectiveness of the proposed approach based on a dataset collected from a famous online health community.FindingsThe experimental results demonstrated that the proposed method outperformed all baseline models in user recommendation using the collected dataset. The incorporation of social information from OHCs can significantly improve the performance of the proposed recommender system.Practical implicationsThis study can help users build valuable social connections efficiently, enhance communication among community members, and potentially contribute to the sustainable prosperity of OHCs.Originality/valueThis study introduces the construction of the UIR network in OHCs by integrating various social information. The conventional MF model is adapted by integrating the constructed UIR network for user recommendation.

show abstract

Friend Recommendation Based on Multi-Social Graph Convolutional Network

Cited by 39 publications

References 28 publications

Multi-Relational Graph Convolution Network for Service Recommendation in Mashup Development

Multi-Relational Graph Convolution Network for Service Recommendation in Mashup Development

SAFER: Social Capital-Based Friend Recommendation to Defend against Phishing Attacks

User recommendation in online health communities using adapted matrix factorization

Contact Info

Product

Resources

About