|本論文探討以寵物作為教育虛擬角色的設計和應用。基於人工智慧和多媒體技術的進步，傳統教育虛擬角色常將電腦模擬成智慧型家教(intelligent tutor)，教導學習者，或程度相近的學習同伴(learning companion)，與學習者一起互動。然而，不論是智慧型家教或學習同伴，大多採用系統主動策略(system-initiative strategy)，電腦掌握主動權，追蹤學習者的認知狀態，並於適當時機介入。本論文則根據學習者主動策略(learner-initiative strategy)，提出動物同伴(animal companion)概念，將電腦模擬成需要飼養和照顧的虛擬寵物，以期有助於學習者與虛擬角色的互動。 飼養寵物是人們生活上常見的普遍現象，尤其是孩童，對寵物似乎特別關愛。若分析孩童和寵物的關係，可發現兩點特徵：主動積極態度、情感依附(attachment)關係。這兩點特徵對學習而言，同樣相當重要。學習者應該被鼓勵採取主動積極的態度，專注於學習任務，並且不斷付出。基於這樣的設計理念，本研究實作了動物同伴系統：我的寵物(My-Pet)和我們的寵物(Our-Pet)系統，並應用於開放學習者模型(open learner model)和遊戲式學習(game-based learning)。 我的寵物(My-Pet)是學生所飼養的虛擬寵物，寵物會有不同的需求，為了滿足寵物的需求，學生必須參與學習活動，並通過所規定的評量活動而賺取金幣，學生即利用這些金幣飼養自己的寵物。我們的寵物(Our-Pet)則是小組所共同飼養的虛擬寵物，由小組內的每個成員一起負責飼養和照顧。 為了克服開放學習者模型的使用動機(usage motivation)和互動性(interactivity)兩項挑戰，My-Pet和Our-Pet扮演主動性鏡子(active mirror)的角色，並應用三項設計策略，以期在學生的動機、反思、和成員互動上，帶來更多助益。此外，My-Pet和Our-Pet基於鹽巴設計觀點(salt design perspective)，採用鬆散結合架構，連結寵物養成遊戲和不同領域的學習活動，透過寵物訓練、競爭等遊戲要素，以悅趣化的學習方式，鼓勵學生不斷努力付出。 在開放學習者模型應用方面，本研究於一個三十一位國小五年級學生的班級中，進行初步的嘗試試驗，以收集學生對我的寵物和我們的寵物 (My-Pet-Our-Pet) 系統、開放學習者模型的反應和回饋。研究結果指出，學生對這樣的開放學習者模型設計，有許多情意面向的正面反應，尤其對My-Pet的關愛和照顧，更甚於Our-Pet。學生為了照顧好My-Pet，有高度的動機願意參與學習活動，並改善自己的學習狀況。 此外，針對遊戲式學習應用方面，本研究採用一個受試者間的實驗設計，以三班共六十八位國小五年級的學生為對象，操作不同版本的我的寵物(My-Pet)系統，進行實驗，用以檢驗數位教材、動物同伴、寵物競爭三者的效用。實驗結果顯示，有動物同伴和寵物競爭的版本，學生的使用動機較數位教材高，尤其是加入寵物競爭的版本，其學習品質(單位時間內的進步幅度)也顯著高於其他兩組。 This study investigates the design of virtual pets as educational virtual characters. In the research field of educational virtual characters, computers are traditionally simulated as intelligent tutors to provide personalized instruction, or as learning companions to provide peer-like interactions based on the technology of artificial intelligence and multimedia technologies. However, most of these systems emphasize the cognitive aspect rather than the affective aspect, and utilize “system-initiative strategy” to monitor the learners’ cognitive status for appropriate interventions. Instead, this study highlights the affective aspect prior to the cognitive aspect, and proposes the concept of animal companion based on the “learner-initiative strategy.” That is, computers are simulated as care-needed pets to motivate learners to learn through game-based learning models. Pet-keeping is a pervasive culture in the human’s life. People, particularly children, seem to have natural attachment to their pets. Analyzing the keeper-to-pets relationship, we could find two apparent characteristics: responsible attitude and attachment relationship. These two characteristics are also crucial to learning. Learners should be encouraged to be responsible for their learning, and make efforts constantly for long period of time. Therefore, based on such rationales, an animal companion system, My-Pet-Our-Pet, is implemented to explore the design and applications: open learner model and game-based learning. In terms of open learner model, to overcome the two challenges of usage motivation and interactivity, animal companions are portrayed as open learner models to benefit children’s learning in motivation, reflection, and member interactions. Furthermore, in terms of game-based learning, a loosely-coupling structure invented by a salt design perspective is proposed. Learning activities are incorporated with game activities with the loosely-coupling way. Contrast to the sugar perspective, the salt perspective means that learning requires constant efforts and sweat. Several game elements, such as pet-nurturing, pet-training, and pet-competition, are embedded in to the learning model. The rationale of such design is to make learning more enjoyable and to encourage students’ effort-making learning behaviors. A trial study was conducted in a 31 fifth-grade pupil classroom for collecting feedbacks and comments on My-Pet-Our-Pet system. A student keeps her own individual animal companion, called My-Pet, which holds the open learner model of the student, and each team has a team animal companion, called Our-Pet, which owns their open group learner model. The results revealed that pupils gave positive affective comments on the portrait of animal companions as open learner model, and were willing to participate in learning activities for taking good care of My-Pet. Nevertheless, the driving force of Our-Pet is not as successful as that of My-Pet. In addition, a between-subjects experiment was conducted among three fifth-grade classes (totally 68 pupils) to examine the effect of three key components in the My-Pet system: digital content, pet-nurturing, and pet-competition. Therefore, three different versions of My-Pet systems were used by three groups. The result showed that the presence of My-Pet is helpful to the pupils’ perception of enjoyable experience. In addition, the complete version (containing digital content, pet-nurturing, and pet-competition) has a better learning quality. That is, the group got more improved score during a shorter period of time.