Pin-Chieh Cheng

This paper proposes a method for recommending recipes considering diversity from the cooking point of view. For recipe recommender systems for health promotion, nutrition is an important factor, and has been considered by many systems. However, as users would only accept recipes they can / want to cook, cooking method should also be considered. As the first step for recommending acceptable recipes for different users, this paper aims to find diversified recipes from the cooking point of view. To achieve the goal, this paper employs four different factors: availability of ingredients, the number of cooking steps, the rareness of cookwares used in the recipe, and the cost to cook it. Diversified list of recipes is obtained on the basis of MMR (maximal marginal relevance), in which dissimilarity function between recipes is defined as JSD (Jensen-Shannon divergence). Preliminary experiment with Rakuten Recipe dataset shows the proposed method can recommended different types of recipes from the cooking point of view.