论文标题

烹饪食谱中的营养概况估计

Nutritional Profile Estimation in Cooking Recipes

论文作者

Kalra, Jushaan, Batra, Devansh, Diwan, Nirav, Bagler, Ganesh

论文摘要

食谱的准确营养概况的可用性是食品数据库的重要特征,其中包括营养援助,推荐系统和饮食分析。通常,在在线数据库中,食谱是从不同来源获得的,以最大程度地提高数据集的配方数量和多种多样。这导致了一套不完整且通常不可靠的营养细节。我们提出了一种可扩展的方法,用于使用其成分部分的营养概况估算,该方法使用标准可靠数据库为营养价值估算。先前的研究已经证明了小数据集上的弦匹配方法的效率。为了证明我们的程序的有效性,我们使用美国农业标准参考部(USDA-SR)数据库中包含来自多个数据源的大型数据集上提出的方法,该方法包含了来自多个数据源的食谱,作为计算营养概况的参考。我们通过计算食谱数据库(每份36卡路里)的平均误差来评估我们的方法,该错误符合物理变化的误差范围。

The availability of an accurate nutrition profile of recipes is an important feature for food databases with several applications including nutritional assistance, recommendation systems, and dietary analytics. Often in online databases, recipes are obtained from diverse sources in an attempt to maximize the number of recipes and variety of the dataset. This leads to an incomplete and often unreliable set of nutritional details. We propose a scalable method for nutritional profile estimation of recipes from their ingredients section using a standard reliable database for the nutritional values. Previous studies have testified the efficiency of string-matching methods on small datasets. To demonstrate the effectiveness of our procedure, we apply the proposed method on a large dataset, RecipeDB, which contains recipes from multiple data sources, using the United States Department of Agriculture Standard Reference (USDA-SR) Database as a reference for computing nutritional profiles. We evaluate our method by calculating the average error across our database of recipes (36 calories per serving) which is well within the range of errors attributable to physical variations.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源