MOBILE-BASED FOOD RECOMMENDATION SYSTEM USING HYBRID FILTERING METHODS

Abstract

In the era of technological progress, information technology-based applications have changed the way we view and choose food. With more and more choices available, many people face the challenge of trying different foods because they are not sure whether a dish will suit their taste. Therefore, a recommendation system was developed to increase user satisfaction and ease of food selection using the Hybrid Filtering method which combines Collaborative Filtering and Content-Based Filtering with the cascade technique. The analysis carried out on this system will be based on food descriptions which include the composition of the food itself such as taste and main ingredients. Meanwhile, users will be asked to fill in preferences based on food descriptions. The operation of this technique involves making recommendations first using the Collaborative Filtering method. Next, the recommendations from this method are further filtered using the Content-Based Filtering method. To calculate the similarity value between items, the cosine similarity formula is used, and to predict missing ratings, the weighted sum formula is used. The research results show that combining these two methods produces better recommendations than using either method separately. The results obtained from the calculation data show that the value produced to find out how big the error value is between the predicted value and the actual value for ten items is 0.29. Apart from that, a survey was also conducted on respondents, and the findings showed that eight out of ten people stated that the recommendations given were in line with their preferences.