近年來資訊技術的快速成長，透過網路與多媒體的結合，可提供具彈性且非線性式學習形式方便學習者可以依個人所需隨時自主學習，因此數位化學習已成為一個重要的學習趨勢。然而，數位化學習的自由、彈性 ，也增加學習者學習的複雜程度，有可能對資訊能力相對弱勢的學習者產生學習上的障礙。因此若能適時加入相關的服務機制設計，呈現最適合的內容和瀏覽支援幫助不同的學習者學習，將可降低資訊複雜度並克服學習者的學習障礙，而適性化服務與環境的提供則是解決此問題的重要機制。其中使用者模式的建立則是適性化環境建構過程中最為重要的考量因素。為能在數位化學習趨勢下提供不同學習者更方便與舒適的學習環境，提高學習者的學習興趣與學習成效，本研究以使用者實際的學習瀏覽行為紀錄，以三種分群技術(k-means、階層式群聚、類神經的ART網路進行群聚分析)，再透過強固群聚分析將此三種分群結果加以整合後對使用者進行分群，並深入探討每一群集中人因變數的相互影響與組合關係，進而建立一穩定之學習者使用模式，作為建構適性化學習系統之基礎。

In recent years, to advances network technology, IT-enabled learning and support learning are important concept in on-line education. More learners obtain knowledge by the Web-learning Instruction (WBI). It is beneficial to provide personalized service to help students learning with the Web-based learning system. To deliver effective personalization of Web-based Learning System (WBLS) users, it is important to provide matching instructional strategies to enhance users’ learning performance. Therefore it is necessary to identify different types of learning model that conducted by users in their learning progresses. This study utilizes k-means, hierarchical clustering and ART clustering technique to understanding users behavior by using eight interaction variables. Moreover, robust clustering, capable of correcting the bias of individual clustering techniques, is used to obtain a deeper understanding. According to the results from k-means clustering, there are three different learning models in WBLS users and users in different level of knowledge and different type of cognitive style using different model to learn.