Graph Neural Network for Customer Engagement Prediction on Social Media Platforms

Date

2021-01-05

Contributor

Advisor

Department

Instructor

Depositor

Speaker

Researcher

Consultant

Interviewer

Narrator

Transcriber

Annotator

Journal Title

Journal ISSN

Volume Title

Publisher

Volume

Number/Issue

Starting Page

4158

Ending Page

Alternative Title

Abstract

Social media platforms such as Twitter and Facebook play a pivotal role in companies’ strategy of engaging customers. How to target potential customers on social media effectively and efficiently is an important yet unsolved question. Predicting customer engagement on social media platforms is facing several challenges that cannot be solved by traditional methods. In this work, we design a framework that leverages individual behavior on Facebook together with network contextual information to predict customer engagement (like/comment/share) of a brand’s posts. We first build a meta-path based Heterogeneous Information Network (HIN) to exploit large-scale content consumption information. We then design a Graph Neural Network (GNN) model combined with attention mechanism to learn structural feature representations of users to make the customer-brand engagement prediction. The proposed model is examined using a large-scale Facebook dataset and the result shows significant performance improvement compared with state-of-the-art baselines. Besides, the effectiveness of attention mechanism reveals the potential interpretability of the proposed model for the prediction results.

Description

Keywords

Crowd-based Platforms, attention mechanism, customer engagement, graph neural network, social media

Citation

Extent

9 pages

Format

Geographic Location

Time Period

Related To

Proceedings of the 54th Hawaii International Conference on System Sciences

Related To (URI)

Table of Contents

Rights

Attribution-NonCommercial-NoDerivatives 4.0 International

Rights Holder

Local Contexts

Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.