Regional Sentimental Flows and Hierarchical Drivers of Urban Waste Sorting: An Integrated OCC-LSTM and LDA-DEMATEL-ISM Approach Using Social Media Data

Effective waste sorting policies are essential for promoting sustainable waste management, yet their impact on regional public sentiments and underlying drivers has received limited attention. This study utilizes a comprehensive framework to analyze how policies shape the regional flow of public sentiments and the hierarchical relationships among influencing factors, using social media data as a lens. This study collected garbage classification data from Weibo, using deep learning model to analyze policy's impact on regional attention flow under different sentiments. It also employed the Latent Dirichlet Allocation (LDA), Decision Making Trial and Evaluation Laboratory (DEMATEL), and Interpretive Structural Modeling (ISM) methods to classify and integrate influencing factors under different sentiment themes. The results indicate that: (1) Positive sentiments spread from the eastern regions driven by policies to other areas, while negative sentiments, though still concentrated in the eastern regions, gradually see increasing influence from policies in central and western regions. (2) Policy support is a fundamental factor in promoting waste sorting. (3) The hierarchical network of influencing factors reveals that policies interact with driving factors at the societal, technological, and individual levels. This study offers comprehensive analysis and support for waste classification policy implementation and public acceptance enhancement.