本研究企圖透過知識本體架構來建立糖尿病領域知識。藉由專家訪談、相關文獻、糖尿病領域的專業網站，如醫療百科全書（MedlinePlus）獲取糖尿病領域相關的知識。之後，先透過Noy, et al.所提出建置知識本體架構之步驟及架構建立一起始糖尿病知識本體論（Ontology），並使用美國史坦福大學生物醫學資訊研究中心所開發之Protége為發展平臺並採用Ontology Web Language之描述性語言來建立與維護整體糖尿病本體論架構。再依據Jiahui, et al.所提出之以「代次」保留本體論架構演化之概念來進行後續糖尿病本體論架構版本之維護。本研究所建立之知識本體及使用者介面並進一步部署於趨勢科技所提供之趨勢雲端服務平臺上，讓使用者如醫護人員、領域專家等相關領域知識社群能在雲端服務平臺環境共同分享及維護知識本體，使此糖尿病知識本體能達到協同式建構、知識分享與應用。本研究所發展之系統並透過給予醫護專業人員實際操作和操作後之使用者介面滿意度問卷調查。藉由知識本體中所包含資訊之評估指標，即類別總數和繼承豐富度，由繼承豐富度可以表現出其知識本體所能表現出知識的能力是較趨於嚴謹還是較趨於一般化的。從實驗結果顯示，獨立建構的知識本體成長空間有限，其所表現出知識的能力遠不及協同作業建構所產生的結果。從協同作業建構的實驗顯示繼承豐富度並不會僅因為類別總數的爆增而持續增長。例如協同作業實驗中，實驗之類別總數為67，而繼承豐富度為4.029，但實驗類別總數為98，而繼承豐富度為3.899，可以發現，知識本體的知識表現呈現越來越嚴謹的情況。由此，本研究最後得以實證協同作業建構知識本體的效果會比獨力完成的效果好且更加具可用性及可信性。

The study of this paper aims to establish an ontology for domain knowledge of diabetes. The knowledge of diabetes is acquired through the interview from experts, the review of relevant literatures and professional Web sites about diabetes such as MedlinePlus. Firstly the steps and framework of constructing knowledge ontology proposed by Noy, et al. is referred to establish an initial ontology of diabetes domain knowledge. Secondly, the Protége Ontology developing platform and Ontology Web Language is adopted to construct and maintain the overall diabetes ontology. Furthermore, the evolution concept of ontology's generations presented by Jiahui, et al.[13] is taken to remain generations and carried on the subsequent maintenance of diabetes ontology. The ontology and the user interface designed in this study is deployed to the Trend cloud services platform which is provided by the Trend Micro. By means of the use of user interface designed in this study, the users such as medical and healthcare professionals may share and collaboratively maintain the ontology. In the end, the system developed by this research was provided to healthcare professional users for practical operation followed by the questionnaire survey about the satisfaction to the system. The evaluation indicators including the number of the classes and the inheritance richness are calculated. These measures describe the distribution of information across different levels of the ontologyinheritance tree or the fan-out of parent classes. The result of the experiment shows that the increase of the size of ontology which is constructed independently is limited and the knowledge representation is nothing near better as the ontology which is constructed collaboratively. The result of the experiment for the collaborative construction of ontology is found that the inheritance richness isn't increased continuously as the growing number of classes. For instances, the number of the classes of the ontology which is constructed by the first subject constructed is 67 while the value of the inheritance richness is 4.029. In contrast, the number of the classes of the ontology which is constructed by the second subject constructed is 98 while the value of the inheritance richness is 3.899. This result might reflect a vigorousness of knowledge that an ontology represents. We conclude that the collaborative construction of ontology for diabetes knowledge is better than the independent construction with regards of availability and reliability of system performance.