Exploring Green Travel Influences on Social Media Using Sina Weibo Data and Machine Learning

Abstract

Green travel can significantly alleviate urban traffic congestion, improve overall travel efficiency, reduce carbon emissions, and improve air quality, so it is necessary to explore the comprehensive green travel influencing factors and formulate targeted green travel strategies to encourage residents to travel green. However, most current studies are based on existing literature and questionnaire surveys, and there is a lack of relatively objective multi-source datasets. This article uses multi-source heterogeneous data to explore the influencing factor system and driving mechanism of green travel behavior intention (GTBI). Based on 261,116 social media data, the Word2vec, Bidirectional Encoder Representations from Transformers, linear discriminant analysis, and uniform mobility approximation and projection algorithms were used for data filtering, identifying GTBI grades, extracting influencing factors, and visualization. Then, regression and association rule algorithms verified the validity of the influencing factors and the correlations between specific influencing factors. Finally, a comprehensive theoretical model of GTBI influencing factors was constructed by combining classical behavioral theory and literature data, and the dual validation of GTBI driving mechanisms was achieved through structural equation modeling.