以食物過敏因子相似度檢測提升健康餐飲的安全性

Improving the Safety of Healthy Catering with Food Allergy Factor Similarity Testing

王宏軒(HONG-HSUAN WANG)

指導教授 ： 王淑卿

朝陽科技大學/資訊學院/資訊管理系/碩士(2020年)

摘要

食物過敏造成的過敏症狀輕則皮膚之搔癢，嚴重者可能會造成造成休克，根據ACAAI(American College of Allergy, Asthma, and Immunology)的數據統計，過敏已經成為全球第六大的疾病。WAO(World Allergy Organization)世界過敏組織在2005年6月28日時將每年的7月8日訂為世界過敏性疾病日。換言之，因過敏而產生疾病是必須被重視的。在現今的社會，人們越來越重視健康，吃著許多昂貴或美味的食物，看似健康，看似是適合自己的食物，但孰不知這些食物可能都蘊含著可能會導致我們有皮膚之癢或生命上的危險的過敏因子，為了可以讓人們可以安心的食用眼前的料理，並且不會有生命上的問題，本研究提出「基於過敏因子相似度檢測的食物推薦系統(Allergy Similarity Detection based Food Recommendation System，ASDFRS)」，根據ASDFRS中的「不建議料理」、「推薦料理」與「查詢料理」可以篩選掉不可食用的料理，也可以推薦可食用的料理。
本研究的食譜資料集是透過網路爬蟲cookpad食譜共享網站抓取「嬰兒」、「幼兒」與「兒童」，食譜的資料集有612筆的資料，在個人資料集中，0-11歲的資料有329,656筆的資料，其中包含有過敏因子的人與沒有任何過敏因子的人。本研究提出的ASDFRS主要是透過食譜資料及與已知食用過敏因子和未知食用過敏因子來進行「不建議料理」、「推薦料理」與「查詢料理」三個機制，而未知食用過敏因子是根據「Microsoft Excel分析個人資料集的結果」與「相關食物過敏的文獻」推論出的結果。在三個機制中都會使用的餘弦相似度(Cosine Similarity)來協助我們判斷食用者對該道料理是否可食用，當Cosine Similarity所算出來的值大於0，代表食用者不建議食用該料理，若Cosine Similarity等於0，代表食用者可以安心的食用該道料理。

關鍵字

Microsoft Excel ； 網路爬蟲 ； 過敏因子 ； 食物過敏 ； 餘弦相似度

並列摘要

Allergic symptoms caused by food allergies range from itching to the skin, and severe cases may cause shock. According to statistics from ACAAI (American College of Allergy, Asthma, and Immunology), allergy has become the sixth largest disease in the world. On June 28, 2005, WAO (World Allergy Organization) designated July 8 as World Allergy Disease Day every year. In other words, diseases caused by allergies must be taken seriously. In today’s society, people are paying more and more attention to health, eating many expensive or delicious foods that seem to be healthy or suitable for themselves, but these foods may contain skin itch or life risk allergic factor. In order to allow people to eat the food in front of them without worrying about life, the Allergy Similarity Detection based Food Recommendation System (ASDFRS) is proposed in this study. According to the “non-recommended dishes”, “recommended dishes”, and “query dishes” in ASDFRS, people can filter out inedible dishes, and you can also recommend edible dishes.
The recipe data set of this research is to crawl “babies”, “toddlers” and “children” through the web crawler cookpad recipe sharing website. And, the recipe data set has 612 data. In the personal data set, there are 329,656 data from 0-11 years old, which contains people with allergic factors and people without any allergic factors. The ASDFRS proposed in this study mainly uses the recipe information and the known eating allergic factors and unknown eating allergic factors to carry out the three mechanisms of “non-recommended dishes”, “recommended dishes”, and “query dishes”. The unknown food allergy factor is the result of inferences based on “Microsoft Excel analysis of personal data sets” and “related food allergy literature.” Cosine Similarity, which is used in all three mechanisms, helps us to determine whether the queried person is edible for the dish. When the value calculated by Cosine Similarity is greater than 0, it means that the queried person does not recommend eating the dish. If Cosine Similarity is equal to 0 means that the person being queried can eat the dish with peace of mind.

並列關鍵字

Cosine Similarity ； Microsoft Excel ； Web Crawlers ； Allergic Factors ； Food allergies

參考文獻

[1] 李志輝、羅平，SPSS常用統計分析教程，2015，北京，電子工業出版社。

[2] 陳榮靜、江慧琴、羅育文(2014)，「一個新的糖尿病患者飲食推薦方法，資訊科技國際期刊」，第8卷，第2期，第22頁-34頁。

[3] 張之妍、 郭啟文、林建亨(2016)，「兒童過敏原檢測流行病學研究：一區域教學醫院之報告」，澄清醫護管理雜誌，第12卷，第1期，第13頁-21頁。

[4] S. Craw (2017) Manhattan Distance. In: Sammut C., Webb G.I. (eds) Encyclopedia of Machine Learning and Data Mining. Springer, Boston, MA

[5] R. S. Gupta, E. E. Springston, M. R. Warrier, B. Smith, R. Kumar, J. Pongracic, and J. L. Holl (2011), “The Prevalence, and Distribution of Childhood Food Allergy in the United States,” Pediatrics, Vol. 128, No. 1, pp. e9-e17.