本研究利用文字探勘技術(text mining),協助專家發展「公司治理」領域的知識本體(ontology)。近年來,電腦系統為達到資訊共用及資源共享之目的,發展各領域的本體已成為知識分享的重要方法;然而,建置各專業領域的知識本體,在實務上仍有賴各領域自行籌組,應用效益因而受限。由於本體建置過程是一項知識密集的作業,須依賴專家提供領域內的經驗或知識,再經由分析、歸納、抽象化及修正等冗長程序,因此建置本體是一項費時耗力的工作。為解決人工作業的瓶頸,本研究以A-G方法為基礎,提出改良的本體建置程序,稱為AGOE(A-G ontology engineering)方法,嘗試由萃取文件中的知識元素著手,整合文字探勘、語意分析等技術,快速建置特定領域的「雛型本體」,以利提供人類專家作後續修改。本研究以「公司治理」領域為例,分別利用文字分析建構詞彚表、建構語意網路、再以語意網路進行分析,逐步完成公司治理領域之本體。由實證的評估顯示,五位領域專家對於本研究產出的雛型本體之同意比例皆甚高,對於需要文字探勘技術輔助來建立本體知識的專業領域,AGOE方法應可作為本體建置前置處理之參考方法。
This study utilizes text mining to assist domain experts in building a prototype of corporate governance ontology. In recent years, ontology becomes an emerging approach in building expertise to achieve reusable and sharable knowledge. However, developing a specific ontology is uneasy due to the development of labor-intensive and time consuming issues. The development process is highly knowledge intensive procedures including analysis, summary, abstraction and revision. To address the above difficulties, this study develops a method, called A-G ontology engineering (AGOE), improved from the A-G method. The AGOE utilizes text mining and related mechanisms for extracting knowledge items and relations into a prototype ontology. Domain experts then take advantages of the prototype to revise for a better ontology. An example of the ”corporate governance” domain is implemented for demonstrating the utilizations of the AGOE method. Main works include creating a glossary, building the semantic network and analyzing their semantics for deriving corporate governance ontology. The empirical evaluation indicates that the proportion of consenting to the prototype ontology from 5 domain experts is very high. Consequently, the AGOE is proved to facilitate users to design ontology in the early development stage, especially for those domains requiring text mining techniques to accomplish the ontology design.