Discovering public attitudes and emotions toward educational robots through online reviews: a comparative analysis of Weibo and Twitter

Abstract

Purpose

Public reviews on educational robots are of great importance for the design, development and management of the most advanced robots with an educational purpose. This study explores the public attitudes and emotions toward educational robots through online reviews on Weibo and Twitter by using text mining methods.

Design/methodology/approach

Our study applied topic modeling to reveal latent topics about educational robots through online reviews on Weibo and Twitter. The similarities and differences in preferences for educational robots among public on different platforms were analyzed. An enhanced sentiment classification model based on three-way decision was designed to evaluate the public emotions about educational robots.

Findings

For Weibo users, positive topics tend to the characteristics, functions and globalization of educational robots. In contrast, negative topics are professional quality, social crisis and emotion experience. For Twitter users, positive topics are education curricula, social interaction and education supporting. The negative topics are teaching ability, humanistic care and emotion experience. The proposed sentiment classification model combines the advantages of deep learning and traditional machine learning, which improves the classification performance with the help of the three-way decision. The experiments show that the performance of the proposed sentiment classification model is better than other six well-known models.

Originality/value

Different from previous studies about attitudes analysis of educational robots, our study enriched this research field in the perspective of data-driven. Our findings also provide reliable insights and tools for the design, development and management of educational robots, which is of great significance for facilitating artificial intelligence in education.