網路的發展使得資料量愈來愈龐大，使用者要從大量的資訊中找到目標是非常困難的，因此本研究利用自建的ontology架構來建立user profile ontology來幫助記錄使用者興趣。本研究主要分兩大部份：自動建立ontology與user profile建立。在自動建立ontology部份，過去大部份的ontology都由領域專家手動建立，非常耗時與維護更新不易，因此本研究提出使用動態文件階層分群來建立ontology架構，其中我們針對部份流程進行優化，改善其準確度與效能。User profile部份，本研究利用自建的ontology架構再加上使用者接觸過的文章與行為因子來建立explicit profile與implicit profile，並且此user profile可以表達使用者不同時期的興趣與達到使用者興趣的延伸，可以幫助使用者找到符合興趣的文章。 The development of Internet makes the amount of data get larger. It is very difficult for user to find the data they want among the huge amount of data. This study creates user profile ontology by using the self-built ontology to record user’s interest. This study consists of two parts: automatically constructing ontology and user profile creation. In the first part, most ontology was created manually by experts, it was not only time-consuming but also hard to maintain and update. Therefore, this study constructs ontology by dynamically hierarchical clustering on document, and we optimize parts of the process to improve the accuracy and performance. In the part of user profile creation, this study use the self-built ontology, the articles that user had read and behavioral factors to create explicit profile and implicit profile. And these user profiles represent the interests in the different periods or the extension of interests of user, it help user to find the article what they may have interest.
