ReciPicker: Recipe recommendation using hybrid filtering for reducing food waste

• Published in: Frontiers in Artificial Intelligence and Applications
• Main Author: Mohd Rosli M.; Abdullah N.A.S.; Ruzaini A.
• Format: Conference paper
• Language: English
• Published: IOS Press BV 2019
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85082045032&doi=10.3233%2fFAIA190070&partnerID=40&md5=83bc25836e73afb3ba6212063483dc3f

In the literature, several researches on recipe recommendation have been proposed, taking into account different goals, such as to suggest healthy recipe and to personalize user taste. However, recommending recipes with a specific goal of reducing food waste is still an open issue. Meal preparation require recipes and ingredients. In some recipes, it is not always clear the quantity of ingredients has been used. If the ingredients are not fully utilise, it will lead to food waste. The goal of this study is to determine a useful way to recommend recipes using available ingredients in the house to prevent household food waste. We construct a hybrid filtering approach that combines the Term Frequency (TF) and Inverse Document Frequency (IDF) as content-based filtering and matrix factorization as collaborative filtering. We develop a recipe recommender system using the hybrid filtering approach. We conduct a preliminary user study to evaluate the usability of the recommender system. We found that all participants agreed that the system is useful and 80% of participants agreed that the system meet their needs. The recipe recommender system can be used to search recipes based on available ingredients. Furthermore, the system will be able to assist user to manage their time by reducing the time taken to prepare meals. © 2019 The authors and IOS Press. All rights reserved.