With intensifying competition in the tourism industry, refined analysis of visitor experience has become increasingly critical for effective destination management. Taking the Chengdu Research Base of Giant Panda Breeding as a case study, this research leverages visitor reviews from online travel platforms to explore emotional tendencies through text mining and sentiment analysis techniques. Using a self-developed Python web crawler, approximately 2,500 valid reviews were accurately collected and rigorously screened from the Ctrip platform. Methodologically, this study innovatively integrates SnowNLP sentiment polarity analysis with Latent Dirichlet Allocation (LDA) topic modeling to systematically examine visitor experience from three perspectives: overall sentiment tendency, evaluation of specific landscape elements, and correlations with management factors. The findings indicate a generally high level of visitor satisfaction, with panda exhibits and natural environments identified as primary sources of positive feedback. Nonetheless, issues such as inadequate infrastructure and suboptimal service quality were also evident. Based on these insights, targeted recommendations are proposed, including optimizing transportation organization, upgrading infrastructure, and enhancing service training to improve overall visitor experience. The methodological integration and detailed analysis provide robust empirical support and practical guidance for tourism management decision-making.