Abstract
Users of social networks have a variety of social statuses and roles. For example, the users of Weibo include celebrities, government officials, and social organizations. At the same time, these users may be senior managers, middle managers, or workers in companies. Previous studies on this topic have mainly focused on using the categorical, textual and topological data of a social network to predict users’ social statuses and roles. However, this cannot fully reflect the overall characteristics of users’ social statuses and roles in a social network. In this paper, we consider what social network structures reflect users’ social statuses and roles since social networks are designed to connect people. Taking an Enron email dataset as an example, we analyzed a preprocessing mechanism used for social network datasets that can extract users’ dynamic behavior features. We further designed a novel social network representation learning algorithm in order to infer users’ social statuses and roles in social networks through the use of an attention and gate mechanism on users’ neighbors. The extensive experimental results gained from four publicly available datasets indicate that our solution achieves an average accuracy improvement of 2% compared with GraphSAGE-Mean, which is the best applicable inductive representation learning method.
Highlights
In recent years, online social networks have become more and more popular in enabling people to connect, communicate, and share information
In order to aggregate the features of the one-hop neighborhood, we introduced a graph neural network (GNN) layer, denoted by Meanlayer
The third method, denoted by GraphSAGE-Mean, applies meanaggregator [4] to iteratively aggregate the information of each hop neighbor without using the gate mechanism
Summary
Online social networks have become more and more popular in enabling people to connect, communicate, and share information. GraphSage uniformly aggregates each node’s neighbors which limits its ability to represent nodes’ features in social networks (e.g., email [7,8,9,10], webpage [11] and citation [12,13,14] networks) It does not take into account the fact that different neighbors’ nodes contribute differently to the representation of the given node. The graph attention network (GAT) [5] specifies different weights to different nodes in a neighborhood by leveraging masked self-attention layers to address the shortcomings of the GCN-based methods, which is applicable to both transductive and inductive problems It cannot differentiate the importance of different levels of a given node’s neighbors. In order to effectively infer users’ social statuses and roles, we propose a novel inferring method that deals with the feature information of users by considering different contributions of a given node’s neighbors.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have