本研究以「明日閱讀習慣暨家校共讀博覽會」認證活動為脈絡，針對學生面對大量書單時不易選書、閱讀內容容易集中於熟悉領域或難度，以及閱讀後整理與回饋不足等問題，設計「書籍地圖」系統。系統整合推薦功能、AI聊書同伴與拼圖任務，分別提供閱讀前選書參考、閱讀中進度與分布視覺化，以及閱讀後反思引導。 本研究採單組前後比較之準實驗設計，並以混合研究法進行分析。研究對象為桃園市某實驗小學參與家校認證活動之33位學生，實驗期程共9週。資料來源包含系統歷程紀錄、閱讀學習動機策略量表前後測與學生半結構式訪談。依資料性質與研究問題不同，本研究分別以推薦配對分析、第一筆與最後一筆聊書比較、完整聊書歷程趨勢分析、上下學期閱讀紀錄比較，以及量表前後測分析，檢視系統導入後學生在選書參考、閱讀反思、閱讀行為與閱讀動機策略上的變化。 研究結果顯示，推薦結果與學生實際閱讀登記具有部分重疊，其中以熱門書籍配對最為明顯，表示推薦功能較適合作為輔助性選書參考。AI聊書同伴透過4F架構，引導學生回顧事實、表達感受、整理發現並連結未來生活；依系統評分結果，學生4F平均分數顯著提升，其中以感受表達與未來連結之提升較為明顯。在拼圖任務機制導入後，本研究觀察到學生有效閱讀登記數量、閱讀廣度與平均閱讀深度皆呈現顯著增加。量表結果顯示，學生自我效能顯著提升，內在價值、自我調節與整體量表未達顯著差異。 整體而言，「書籍地圖」將閱讀認證任務轉化為可視覺化、可追蹤且具回饋性的歷程，並支持學生選書、持續閱讀與閱讀後反思。本研究可作為未來閱讀認證活動推動與AI閱讀輔助系統設計之參考。;This study designed the “Book Map” system for the Tomorrow Reading Habits and Home-School Collaborative Reading Certification Program. The system addresses students’ difficulties in selecting books from large reading lists， their tendency to remain within familiar domains or difficulty levels， and limited post-reading organization and feedback. It integrates recommendation functions， an AI book-talk companion， and puzzle tasks to support book selection， visualize reading progress， and guide post-reading reflection. A single-group pre-post quasi-experimental design with mixed-methods analysis was adopted. Participants were 33 students from an experimental elementary school in Taoyuan City， and the study lasted nine weeks. Data included system logs， pretest and posttest questionnaires， and semi-structured interviews. Analyses included recommendation matching， AI book-talk comparisons and trends， semester-based reading record comparisons， and questionnaire analysis. Results showed that the recommendation functions mainly served as auxiliary book-selection references. Based on system-generated scores， students’ average 4F scores improved significantly， especially in expressing feelings and connecting reading to future life. After puzzle tasks were introduced， valid reading records， reading breadth， and average reading depth increased significantly. Questionnaire results showed significant improvement in self-efficacy， while intrinsic value， self-regulation， and the overall scale showed no significant differences. Overall， the “Book Map” system transformed reading certification tasks into a visualized， trackable， and feedback-oriented process supporting book selection， sustained reading， and post-reading reflection.
