Explaining and Predicting Helpfulness and Funniness of Online Reviews on the Steam Platform

Zhi Wang,Gergely Horvath,Victor Chang

doi:10.4018/jgim.20211101.oa16

Abstract

Online review is a crucial display content of many online shopping platforms and an essential source of product information for consumers. Low-quality reviews often cause inconvenience to the platform and review readers. This article aims to help Steam, one of the largest digital distribution platforms, predict the review helpfulness and funniness. Via Python, 480,000 game reviews related data for 20 games were captured for analysis. This article analyzed the impact of three categories of influencing factors on the usefulness and funniness of game reviews, which are characteristics of review, reviewer and game. Additionally, by using the Random Forest-based classifier, the usefulness of reviews could be accurately predicted, while for funniness, Gradient Boosting Decision Tree was the better choice. This article applied research on the usefulness of reviews to game products and proposed research on the funniness of reviews.

Highlights

Online user reviews have gradually become an important display content of many online shopping platforms with the full application of Web 2.0 and the rise of social networking sites and new media
As mentioned in the methodology part, Poisson and Zero Inflated Poisson regression model is applied to analyze the influence of different dependent variables on review helpfulness and funniness rated by community members in this dissertation
The recommendation has a negative impact on review helpfulness, indicating that reviews that do not recommend this game instead are more informative

Summary

Introduction

Online user reviews have gradually become an important display content of many online shopping platforms with the full application of Web 2.0 and the rise of social networking sites and new media. It refers to consumer’s judgment (evaluation, response, recommendation, or complaint) on a product or service after purchase on the Internet. These reviews contain a large amount of useful information, and because the data comes from different individuals, it can avoid the one-sidedness of information. The amount of consumer reviews has been increasing rapidly, creating significant big data challenges for consumers and businesses (Singh et al, 2017)

Objectives

Methods

Results

Conclusion