Research on Digital Cultural Tourism System Based on LDA Model

In the era of digital transformation, cultural tourism has become increasingly reliant on intelligent systems that can extract, organize, and recommend information effectively to enhance user experiences. However, the vast amount of cultural tourism data, including text from travel blogs, reviews, heritage documents, and online platforms, often suffers from redundancy, sparsity, and lack of semantic organization, making it difficult for users to access meaningful insights. To address these issues, this research implements a Latent Dirichlet Allocation (LDA) model for topic modeling and knowledge discovery within the digital cultural tourism domain. The LDA model identifies hidden thematic structures in unstructured textual data, enabling the classification of tourism resources into interpretable topics such as heritage, food culture, architecture, and local traditions. For experimentation, publicly available datasets such as Cultural Heritage Corpus, Tourism Review Dataset (TripAdvisor/OTA platforms), and WikiTravel Text Data were utilized. The proposed approach significantly improves semantic clustering and recommendation accuracy compared to traditional keyword-based methods. Experimental results demonstrate that the LDA-based system enhances cultural tourism information retrieval with accuracy 95.72%, 94.35% precision, 93.81% recall, 94.92% Specificity and 94.08% f1-score provides personalized travel guidance, and supports the construction of a digital cultural tourism knowledge base, thereby promoting smarter and more engaging cultural tourism experiences.