New developments in culinary computational creativity

Abstract

In this paper, we report developments in the evaluation and generation processes in culinary computational creativity. In particular, we explore the personalization aspect of the quality and novelty assessment of newly created recipes. In addition, we argue that evaluation should be a part of the generation process and propose an optimization-based approach for the recipe creation problem. The experimental results show a more than 41% lift in the objective evaluation metrics when compared to a sampling approach to recipe creation.