Abstract: One of the important applications in modern computing is to provide better travel services for tourists. This paper provides a study of utilizing online travel information for the personalized travel package recommendation. Many recommender systems lack an organized framework to dynamically incorporate multiple types of additional context information existing in the tourism domain, such as the travel area, season, and price of the travel packages. First we analyze the properties of the previous travel packages and develop a TAST (tourist-area-season-topic) model. This TAST model helps to represent different travel packages and different topic distributions of tourist, taking out of topic is specified on both the tourists and the natural characteristics of the landscapes. According to the topic model representation, a cocktail approach is incepted so that to form lists for personalized travel package recommendation. The TAST model is expanded to the tourist-relation-area-season topic (TRAST) model for collecting the relationships among the tourists for all travel groups. Then analyze both the TAST and TRAST models, and cocktail recommendation approach on the current travel package information TAST model works better than the old recommendation system by adding touristís relationships as TAST model can effectively catches the unique properties of travel data and cocktail approach, an effective estimation for travel group formation can be done using TRAST model.

Keywords: TRAST model, K-means clustering, Authentication, Collaborative filtering, Searching Techniques.