DEVELOPMENT OF RECOMMENDER SYSTEM USING HYBRID CONTENT-BASED AND KNOWLEDGE-BASED TECHNIQUE FOR INDONESIA TOURISM EVENT

• Main Author: YULIA SARI (NIM: 23216132), WULAN
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/31683
• Institution: Institut Teknologi Bandung

<p align="justify">Event recommendation is a new research area in recommender system (recsys). Previous research studies events in one particular category and does not cover all tourism event categories, such as: music events, scientific, academic and Event-Based Social Networks (EBSNs). Later, in its development, the event recsys experiences several other problems, i.e the characteristic of the event which is one and only item. An event always takes place in the future, with a short time and a specific place resulting in each event can be different each year. Not necessarily the same event will happen again in the future. Therefore, feedback from users is difficult to obtain. This makes the event recommendation cannot be done by utilizing user ratings. Collaborative filtering techniques (CF) cannot overcome the problem of time-related items as well as events that usually get ratings after the event is completed. Therefore, this study hybrid techniques of content-based (CB) and knowledge-based (KB). The cold-start challenge and the portfolio effect problem resulted by CB techniques can be solved by KB techniques. The objectives of this study are to produce personalized event recommendations and diversity, considering that the system not only meets the needs of users but also as a media for promoting Indonesian tourism events. The results of this study conclude that the incorporation of CB and KB on its application in the case of tourism events is possible and that the two techniques become complementary one another. Experiment showed that this recysys design resulted in a diversity score above 0,6. Then the evaluation results of the user survey indicate the developed design can be accepted by the user based on its usability and satisfactionary aspects, and also considered to be able to provide recommendations that are very interesting for its users. <br /> <p align="justify">