Travel Recommender System for Social Media

Abstract

With the help of personalized recommender systems nowadays the customized services to users are made available. Social media plays an important role to develop a personalized recommender systems. In this study we explores the twitter data to personalize recommendations for the users who travels most. Using a machine learning classification model we identify the travel related tweets. These travel related tweets are then applied to personalize recommendation system to know places of interest for the user. We classified the places of interest such as: Old or historical buildings, gardens, ancient museums , famous places and restaurants. To get the better results we also considered the travel tweets of friends, family members. The model has been evaluated by comparing the predictions made by the model with the users choices in the survey. These evaluations show 70% prediction accuracy. This accuracy can be increased with good travel-tweet training dataset and travel category identification technique using machine learning. In order to refine the recommendations , different categories for travels can be included like sports grounds, musical shows, theaters, etc. The proposed model lists places of users interest from each category in proportion to the travel category score generated by the model.

Keywords: recommender systems, social media analysis, twitter data, personalization, travel tweet