Recommendation Systems for Nutritional Advice: Evaluation of a Content-based Approach

Michal Arkadiusz Blaszczak

Student thesis: Master thesis


Background Problem Previous research related to recommendation systems has described complexity in evaluating the accuracy and performance (Herlocker et. al., 2004). Design choices can vary depending on the nature of items being recommended to the user. When digital platforms adopt recommendation systems, they can rely on network effects (Boudreau & Hagiu, 2009) to retrieve similar users and suggest relevant items. It is however more complex to not suggest less relevant items, and therefore automated techniques should be integrated with more data including certain forms of user profile and preference to further strengthen the suggestions. From the perspective of a user interacting with a content platform it becomes impossible consult a library of at least hundreds of thousands of products, therefore the role of a recommendation system is to improve the navigation experience by highlighting items that might be relevant to the user according to different parameters including similar ones in characteristics of those previously accessed, or items that have been accessed by other users with similar previous search history. The goal of a recommender system is not to suggest items that a user already accessed but to discover new items that might potentially be of interest, this concept is also known as serendipity (Herlocker et. al., 2004). Purpose This study investigates the applicability of information retrieval, in specific content-based recommendation systems, in the context of suggesting a meal plan based on the availability of products in the user’sregional location and specific nutritional preferences. The combination of meal and dietary suggestions and locally available grocery products can potentially reduce food waste, help budgeting, provide a more varied experience by discovering new recipes; the focus is observing the relevance of meal suggestions and their possible applicability throughout a year.

EducationsMSc in Business Administration and Information Systems, (Graduate Programme) Final Thesis
Publication date2021
Number of pages59
SupervisorsWeifang Wu