A Recipe Decision Support System Using Knowledge Information and Agent

Abstract

We report on the development of a recipe recommendation system using collaborative filtering and impression words. As a user interface (UI), we adopted MikuMikuDanceAgent (MMDAgent). In our system, the first recommendation process using collaborative filtering is terminated based on the previously decided condition, after which the second recommendation process identifies the recipe that is most similar to past successful recommendations from recipes that have not been recommended thus far. As a final step, one recipe is selected from all successfully recommended recipes.