Predicting Portuguese Steam Review Helpfulness Using Artificial Neural Networks

• Jardeson L. N. Barbosa UFPI
• Raimundo Santos Moura UFPI
• Roney L. de S. Santos UFPI

Resumo

Online consumer reviews are currently available in most eCommerce websites. By discussing products features, reviews help consumers during the search stage of the buying process. However, from thousands of reviews published daily, it is not always clear which ones consumers should read. To help with that, consumers can vote user reviews as helpful or not helpful. This paper proposes an automatic method to analyze the helpfulness of online consumer reviews. This analysis was made on reviews collected from Steam using Multilayer Perceptron Artificial Neural Network. We found out indication that certain features of reviews affect the perception of helpfulness and we discuss applications and future researches.