NEW LOCATION RECOMMENDATION TECHNIQUE ON NETWORK

Abstract

Nowadays, recommender systems play a crucial role in our daily life as can be seen from numerous websites that utilize the systems to recommend interesting items for their visitors, such as lonelyplanet.com, netfliex.com etc. From operational viewpoint, the recommender system is embedded in several domains by which it is categorized based on their usage. For example, lonelyplanet.com is categorized as a Location-Domain while netflex.com is in a Movie-Domain. In fact, the interesting attractions from traveling websites or movie selections from movie online sites are part of the recommendation outcome. To analyze the user’s preferences, many methods have been widely used including Content-based filtering and Collaborative filtering. Nevertheless, both methods have different benefits and drawbacks. As a result, a hybrid recommender system using more than one technique has been proposed to supplement the short-coming of one technique by the strength of the other technique. Additionally, with the popularity of social network, the new social filtering technique has also been deployed to search for the impact among users on the social network. This thesis proposed a novel method to recommend tourist attractions for the visitors implemented on the location-domain with the combination of 3 recommender system techniques: Content-based filtering, Collaborative filtering, and social-filtering. The comparative results show that the proposed method yields better performance of coverage result and NDCG average score than the other two methods.