Data Mining-Driven Personalized Recommendation in the Smart Tourism Systems

In-depth mining analysis and systematic study of information search behavior are currently lacking in smart tourism and the personalized differences in tourists’ information needs and search methods in different tourism situations have not been fully explored. Information overload, false information and difficulty in information screening are major issues affecting tourists’ decision-making and experience. This paper uses DM (data mining) technology to analyze tourists’ information search behavior in smart tourism, explore the factors that affect tourists’ information search decisions and give optimization suggestions to improve personalized services and information recommendation mechanisms in smart tourism systems. The study cleans and preprocesses the collected data, uses the K-Means clustering algorithm to identify patterns in tourist behavior, divides tourists into different groups and explores the differences in their information needs in various tourism situations. Then, the potential correlation in the process of tourist information search is analyzed through association rule mining (Apriori algorithm) to understand tourists’ preferences and habits in travel decision-making. To achieve personalized recommendations, collaborative filtering and content recommendation models are combined to provide customized information recommendations based on tourists’ historical behaviors and preferences. Finally, the reinforcement learning (Q-learning) algorithm is used to adjust the recommendation system in real time based on tourists’ feedback to improve recommendation accuracy and Visitor Satisfaction (VS). Experimental results show that the personalized recommendation system based on DM is significantly better than the traditional keyword-based recommendation system in terms of precision (0.83), recall (0.79), F1 value (0.81), click-through rate (0.45), conversion rate (0.38) and VS (4.5 points), verifying its advantages in improving tourist experience and decision support. To verify the effectiveness of the optimization strategy, the Information Content Accuracy (CA), Semantic Extraction Precision (SEP) and Response Time (RT) indicators were also tested. The results showed that the optimized system improved the information CA and SEP by 12.3% and 11.8% compared with the original system, the RT was reduced by 50% and the VS was improved from 3.8 points to 4.6 points. In addition, the system has also made breakthroughs in false information filtering, with the false advertising recognition rate increased by 32% and the false comment recognition rate increased by 28%, greatly improving the relevance and practicality of information. The experimental results verify the advantages of personalized recommendation systems based on DM technology in smart tourism, which can effectively improve tourists’ experience and decision-making quality.