Research on Tourists' Sentiment Tendency to Scenic Spots Based on the Transformer Deep Learning Model: A Case Study of Jade Dragon Snow Mountain

The rapid development of mobile internet and digital technology has brought significant changes to the tourism industry, closely linked to shifts in people’s travel preferences. Intelligent analysis of the sentiments conveyed in tourists' comments on social media has become crucial in understanding public demand for tourist attractions. However, existing sentiment analysis methods often lack consideration for the contextual relationships within text paragraphs, leading to inaccuracies in sentiment analysis. This study focuses on Transformer models in deep learning, introducing self-attention mechanisms and designing positional encodings for text segmentation to better account for contextual connections in long texts. It combines traditional dictionary annotation, clustering analysis, and keyword extraction methods as supplementary techniques to enhance the accuracy of sentiment analysis. Experimental results demonstrate that the approach based on Transformer model predictions, combined with traditional dictionary annotation, contributes to capturing the emotional polarity of visitors to tourist attractions. Keywords extracted through clustering accurately capture semantic information and positive/negative emotions in tourist comments, providing researchers in related fields with a theoretical foundation and practical reference. This approach also offers reliable data and a scientific basis for decision-making across various components of the tourism industry.