텍스트 분석 기법과 식재료 계층구조를 활용한 음식 레시피 추천 방법

Abstract

Even in the age of digital technology, food is still one of the most important part of everyday life for all of us. Wecan also easily access food-relevant data about the topic on social media, from trends like diet issues to variousculinary recipes. In this study, we performed the text analytics on the natural language recipes, and proposed therecipe recommendation method considering our preference and ingredients at hand. The ingredient componentswere extracted from the natural language recipes using the Conditional Random Fields, and the popular orunexpected ingredient match was explored using the modified Pointwise Mutual Information and InverseDocument Frequency techniques. Finally, we provided the insights to design new recipe considering alternativeways in case we have insufficient ingredients, which is inferred by the statistical and linguistic analysis.