基於領域本體結合決策樹之慢性病營養素飲食推薦

The Nutrients of Chronic Diet Recommended Based on Domain Ontology and Decision Tree

丁昱銜(Yu-Hsien Ting)

指導教授 ： 陳榮靜 ； 陳靖國

朝陽科技大學/資訊學院/資訊管理系/碩士(2015年)

摘要

近年來資訊的發達，科技的進步，人們開始關心飲食對健康的重要性。現今高齡社會的到來以及現代人的生活型態、生活不規律、長期不健康的飲食、工作壓力大等因素，導致慢性疾病的纏身，如糖尿病，高血壓及高血脂等等。然而大多數的飲食推薦系統，無法針對三高患者給予個人的飲食推薦，且固然吃對食物但未規劃營養素攝取是否平均。因此，本研究將針對三高患者建置一個具有專家知識背景的營養素飲食推薦系統，給予三高患者便利又更精確的飲食推薦。本研究採用本體論(Ontology)、決策樹及Jena來建置本系統。本體論是使用Protégé工具來建立食物知識本體及個人知識本體的知識庫；決策樹是採用Weka資料探勘與機器學習的工具，分析使用者的個人資訊並分類到符合的營養素群組；最後使用Jena的推論引擎，經由本體的關聯性與Jena的規則，篩選不適合使用者食用的食物，並給予符合使用者偏好的飲食推薦。本研究經由以上方法，實作出一個基於領域本體結合決策樹之三高患者營養素的飲食推薦。本系統的飲食推薦結果，透過營養師的評估，並經過精確度的驗證，實驗結果達到100%。因此本系統的飲食推薦，可以提供符合三高慢性病患者營養素的飲食推薦，以達到便利又健康的飲食。

關鍵字

飲食推薦 ； 營養素 ； 慢性疾病 ； 決策樹 ； 本體論

並列摘要

In recent years, due to the development and advance in information science and technology, people care more about the healthy diet, so diet types gradually change and take more focus on health management. Nowadays, it is becoming an aging society in Taiwan, people have irregular life, long-term unhealthy diet, work pressure and other factors to be chronic disease, such as diabetes, hypertension and high cholesterol, and so on. However, most of the dietary recommendation system cannot give the dietary recommendations for patients of chronic diseases. The patients care what food are edible, but they do not notify whether the nutrients are in balance. Therefore, this study built a diet recommendation system of chronic diseases with an expert knowledge, and gave chronic diseases more convenient and more precise dietary recommendations. In this study, we use ontology, decision tree and Jena to build the recommendation system. Ontology is using Protégé tool to build the knowledge of food and personal ontology. Then decision tree is setup by Weka tool which uses data mining and machine learning. The Weka is used to analyze the user's personal information to classify and to match the nutrients groups. Finally, the system uses Jena inference engine, via ontologies relationship and Jena rules to filter not suitable food for users, and give dietary recommendations in line with user preferences. In this study, we design the nutrients of chronic diet which is recommended based on domain ontology and decision tree. The dietary recommendations result is through assessment of dietitians, and verification accuracy is 100%. Therefore, this system of dietary recommendations can provide dietary recommendation of nutrient for patients of chronic diseases to achieve convenient and healthy diet.

並列關鍵字

Nutrients ； Chronic ； Decision Tree ； Ontology ； Dietary Recommendation

參考文獻

[40] 許惠恆 (2000)，「糖尿病的飲食保健療法建議」，中華民國內分泌暨糖尿病學會會訊，第十三卷，第二期，第 3-7頁。

[36] 彭巧珍 (2002)，「肥胖症的飲食治療」，台灣醫學，第六卷，第一期，第46-51頁。

[2] Chiu-Ming Hu (2006), “A nutrition analysis system based on recipe ontology,” University of Taipei Medical.

[38] 林淑媛 (1998)，「尿結石之飲食保健療法與護理指導」，護理雜誌，第四十五卷，第六期，第81-86頁。

[18] Martin Svensson, Kristina Höök, and Rickard Cöster (2005), “Designing and evaluating kalas: A social navigation system for food recipes,” ACM Transactions on Computer-Human Interaction (TOCHI), Vol. 12, No. 3, pp. 374-400.