Recommendation of places of interest for tourists from geo-tagged data using machine learning approaches

With the advanced technologies like global positioning system (GPS), personalised travel recommendations can be done considering user preferences of places by collecting information from diverse sources. Social media websites like Twitter and Facebook can be leveraged to extract geo-tagged data. Geographical location of the user reflects the user preference and can be semantically categorised based on its type using four-square classifier. For identifying similar users preferences of location, collaborative filtering is applied in this work. User preferences are identified from users visit histories and extracted preferences are represented in form of sequences. The frequent sequences are mined using prefix-span algorithm. Sequential rules are mined by applying CM-rules algorithm. By knowing the sequential rules, the subsequent place of interest which is nearer to the user's current location is recommended using k-nearest neighbour (kNN) algorithm. The proposed system has been implemented and the system is found to be promising in terms of accuracy and execution time.