A travel package contains POIs of various categories (e.g., food, attractions, and entertainment), but also the visiting sequence of the selected POIs. These POIs can belong to multiple categories and their attraction may vary with time.
In this paper we focus mainly on recommending travel packages to a user group exploring a particular region, where existing systems lacking. It shows similar personalization, if not better, on single user case also. It can be expanded to long distance travel scenarios by considering whole travel as connecting a series of travel regions. Our system works well even when data availability about a region is less due to variety of metrics used.
In our system we first construct a user profile and location model from the location based social data and find regular sequences among locations. User group and tour details are taken as input from the user. We then find nearby POIs in region and find their likeness by using user interests. Overall group preferences are obtained by giving equal weight to all users but due to considering social connections gives more priority to popular ones opinion .Variety of used metrics provides importance to even less popular but strong user opinions. Popular POIs of each time period are short listed and all possible routes among them which not contradicting with regular sequences are considered. Each route is evaluated by finding transition probability from one location to another and ranked accordingly. Finally top ranked packages are recommended to the user.