Dynamic Topic Evolution and Sentiment Interaction in Pacific Rim Ecological Discourse

This study investigates the dynamic characteristics of ecological discourse in Pacific Rim news and its impact on international ecosystems by integrating Latent Dirichlet Allocation (LDA) topic modeling and sentiment lexicon analysis. A time-sensitive LDA model and a hierarchical weighted sentiment analysis framework were developed to analyze English news texts from 2021 to 2025, identifying five core themes and their sentiment evolution patterns. The findings reveal a dual tension between collaborative symbiosis and competitive confrontation in ecological discourse. Positive sentiments dominate educational and economic cooperation, while negative sentiments correlate with geopolitical conflicts. Cross-evolution analysis demonstrates that discourse fluctuations stem from the interplay between the pluralistic harmony philosophy and power competition realities, highlighting structural contradictions in policy-practice decoupling and discourse power imbalance. Methodologically, this research innovates by proposing a dynamic topic-sentiment evolution framework, enhancing LDA interpretability through temporal parameters and domain-specific lexicon constraints, and bridging ecolinguistics with computational linguistics. The results offer data-driven insights for global ecological governance and advance interdisciplinary discourse analysis.