Based on collaborative filtering algorithm (CFA), this study aims to construct a set of tourism destination recommendation and marketing models. In the context of the rapid development of tourism and the advancement of information technology, personalised travel recommendation has become an important issue of concern in the industry. The study employs a collaborative filtering algorithm to classify users into similar groups by analysing their historical behaviours and preferences, and to provide them with highly personalised destination recommendations. Specifically, the algorithm discovers user groups with similar interests by clustering users. Then, based on the historical rating data of the user groups, the system predicts each user's level of interest in unknown destinations and recommends destinations that may be of interest to the user. The study successfully developed a data-driven destination recommendation and marketing optimisation system by integrating multi-source data through the use of collaborative filtering algorithms. The system provides personalised destination recommendations through in-depth analysis of user behaviour, thereby enhancing user experience and service quality. The approach of integrating multi-source data supports the comprehensiveness and accuracy of the recommendation system, and provides a more effective means for marketing efforts in the tourism industry.