臺灣博碩士論文加值系統

現今軟體人才需求逐年成長，薪資條件也相對優渥，吸引許多跨領域人才投入軟體技能的學習。然而，參加實體課程常有空間與時間的限制，因此線上影音課程日益受到跨領域學習者的喜愛。經營一門優質的線上學習課程不僅包含錄製及剪輯，修習線上課程的學習者人數往往遠大於傳統課程的修習人數，而學習者在學習過程中遇到的的問題通常是由教師或輔導者個別進行人工處理來替學習者解答，後續需要投入大量的人力及時間回答學員的問題。如何降低重複性常見問題，減少輔導者及教師的負擔成為一個非常值得探討的議題。熱門線上課程的問答可能高至幾千則，以人工評估重複性問題太過於耗時，如果將這些問答透過文字探勘技術整理及分析，回饋給課程開發團隊做為調整課程內容的依據，可望有效降低回答學員學習問題的工作量，並能提高學習者的學習滿意度。因此，本研究的目的與內容即是以國內某大付費學習平台中問答數量最多的前端程式設計課程為例，運用文字探勘技術，對該線上影音課程問答集進行詞雲、聚類、主題等分析，找出課程學習過程中學生最常有疑問的課程內容，依此分析結果，課程開發團隊進而整相應的課程內容。根據實地驗證結果顯示，本研究的方法可顯著降低重複性常見問題、減少輔導者及教師的負擔，並提升學習者學習成效。

The demand for software developers has increasedgradually. High salary attracts lots of cross-disciplinary learners to studysoftware skills. However, traditional classes often limited by space-timeconstraints, so online audio-visual courses have become more popular forcross-disciplinary learners. It takes more than just recording and editing tooperate a high-quality online learning course. The learners of online coursesare often much more than traditional in-person courses. Each problem asked bylearners during their learning process is usually answered by teachers orteaching assistants in person. However, it requires a lot of manpower and timeto answer those questions. How to decrease repeating problems in order toreduce the burden of teachers and teaching assistants has become a topic thatworth our discussion.A popular online course may include thousands of questionsand answers. It is not efficient to evaluate repetitive questions manually. Ifall of these questions and answers can be classified and analyzed throughtext-mining technology, feedback given to the course development team as areference to revise the course content is expected to effectively reduce theworkload of teachers and improve user’s experience. The purpose of thisresearch is to take a front-end web development course that has the mostquestions and answers in a domestic paid learning platform as an example. Byperforming text-mining technology to analyze word cloud, text clustering, andtopics of the online audio-visual course question and answer collection. Weanalyze and find out the questions that students asked most frequently duringlearning. Then the course development team can revise the course content basedon this analysis result. The result of this research reveals that our methodcan significantly decrease the repetitive common problems, reduce the burden ofthe teachers and teaching assistants, and improve the learning effectiveness ofthe learners.

目錄
第一章 緒論..................................................................................................................1
    第一節 研究動機............................................................................................1
        壹、 系統性學習前端程式逐年受到重視............................................1
        貳、 降低線上課程中的重複性常見問題............................................4
    第二節 研究目的............................................................................................5
    第三節 論文架構............................................................................................6
第二章 文獻回顧..........................................................................................................7
    第一節 有效學習程式設計............................................................................7
    第二節 線上課程的問題與挑戰....................................................................8
    第三節 文字探勘技術應用............................................................................9
第三章 研究方法........................................................................................................10
    第一節 研究與開發架構..............................................................................10
        壹、 研究架構......................................................................................10
        貳、 開發架構......................................................................................11
    第二節 研究材料..........................................................................................12
        壹、 課程主題 VueJS ..........................................................................14
    第三節 研究工具..........................................................................................14
        壹、 程式語言......................................................................................14
        貳、 分詞工具......................................................................................15
        參、 數據分析......................................................................................17
    第四節 資料前處理......................................................................................18
        壹、 資料清理......................................................................................18
        貳、 分詞處理......................................................................................22
    第五節 資料分析..........................................................................................25
        壹、 文件向量......................................................................................25
        貳、 關鍵字提取..................................................................................26
        參、 主題模型......................................................................................27
        肆、 文本聚類分析..............................................................................28
    第六節 實驗設計..........................................................................................29
第四章 研究結果與討論............................................................................................31
    第一節 資料清理結果..................................................................................31
    第二節 關鍵字分析結果..............................................................................32
        壹、 各章節關鍵字討論......................................................................................35
    第三節 主題建模結果..................................................................................39
        壹、 困惑度計算結果..........................................................................40
        貳、 主題命名及建模結果..................................................................40
    第四節 文本聚類結果..................................................................................51
        壹、 間隔統計結果..............................................................................51
        貳、 各聚類關鍵字結果與討論..........................................................51
    第五節 實驗結果..........................................................................................61
第五章 結論與建議....................................................................................................65
    第一節 研究結論..........................................................................................65
    第二節 研究貢獻..........................................................................................65
        壹、 理論貢獻......................................................................................66
        貳、 實務貢獻......................................................................................66
    第三節 研究限制與建議..............................................................................67
參考文獻......................................................................................................................68
    中文文獻..............................................................................................................68
    英文文獻..............................................................................................................69
附錄..............................................................................................................................72
    課程補充前章節及講座列表..............................................................................72
    Unicode 全形與半形字符對應表.......................................................................77
    問答截圖範例......................................................................................................79


圖目錄
圖 1-1-1. GitHub 年度 Octoverse 報告....................................................................2
圖 1-1-2. CakeResume 熱門程式語言台灣月薪職缺............................................3
圖 1-1-3. 104 人力銀行於 2021 年 3 月更新之熱門程式語言職缺......................3
圖 3-1-1. 研究架構圖............................................................................................11
圖 3-1-2. 研究開發流程架構圖............................................................................12
圖 3-4-1. 資料清理流程圖....................................................................................18
圖 3-4-2. 字串全形轉半形方法............................................................................19
圖 3-4-3. 原始數據狀況圖....................................................................................20
圖 3-4-4. 清理後數據狀況圖................................................................................21
圖 3-4-5. 章節資料清理結果狀況圖....................................................................22
圖 3-4-6. 分詞處理開發流程圖............................................................................23
圖 3-5-1. LDA 機率模型圖....................................................................................27
圖 3-6-1. 實驗設計圖............................................................................................30
圖 4-2-1. 全文本 TF-IDF 分析結果文字雲 .........................................................32
圖 4-2-2. 章節 1 關鍵字文字雲............................................................................35
圖 4-2-3. 章節 4 關鍵字文字雲............................................................................36
圖 4-2-4. 章節 6 關鍵字文字雲............................................................................37
圖 4-2-5. 章節 9 關鍵字文字雲............................................................................37
圖 4-2-6. 章節 11 關鍵字文字雲..........................................................................38
圖 4-2-7. 章節 12 關鍵字文字雲..........................................................................39
圖 4-3-1. LDA 困惑度最優運算結果....................................................................40
圖 4-3-2. LDA 主題間距圖....................................................................................41
圖 4-3-3. 主題 1 前 30 最相關術語......................................................................43
圖 4-3-4. 主題 2 前 30 最相關術語......................................................................44
圖 4-3-5. 主題 3 前 30 最相關術語......................................................................45
圖 4-3-6. 主題 4 前 30 最相關術語......................................................................46
圖 4-3-7. 主題 5 前 30 最相關術語......................................................................47
圖 4-3-8. 主題 6 前 30 最相關術語......................................................................48
圖 4-3-9. 主題 7 前 30 最相關術語......................................................................49
圖 4-3-10. 主題 8 前 30 最相關術語....................................................................50
圖 4-4-1. Gap Statistic 分群群數分析結果 ..........................................................51
圖 4-4-2. 聚類 1 關鍵字文字雲............................................................................54
圖 4-4-3. 聚類 2 關鍵字文字雲............................................................................55
圖 4-4-4. 聚類 3 關鍵字文字雲............................................................................56
圖 4-4-5. 聚類 7 關鍵字文字雲............................................................................57
圖 4-4-6. 聚類 10 關鍵字文字雲..........................................................................59


表目錄
表 3-2-1. 課程補充前章節 11 講座列表..............................................................13
表 3-3-1. 分詞詞性說明........................................................................................15
表 3-4-1. 文本分詞後詞性及出現次數................................................................24
表 3-5-1. 文件向量結構範例................................................................................25
表 4-1-1. 章節編號及發問數量............................................................................31
表 4-2-2. 各章節關鍵字........................................................................................32
表 4-3-3. LDA 主題命名結果...............................................................................42
表 4-4-1. 各聚類關鍵字分析結果........................................................................52
表 4-4-2. 各章節聚類數量....................................................................................60
表 4-5-1. 課程補充後章節 11 講座列表..............................................................61
表 4-5-2. 註冊人數及發問數量統計....................................................................63
表 4-5-3. 原版課程及加強版課程章節 11 關鍵字..............................................63
表 4-5-4. 作業通過狀況統計................................................................................64

參考文獻


中文文獻
104人力銀行。取自: https://www.104.com.tw, [造訪: 21-Mar-2021]
Cakeresume。取自: https://www.cakeresume.com, [造訪: 21-Mar-2021]
Jieba結巴中文分詞。取自: https://github.com/fxsjy/jieba, [造訪: 02-Apr-2021].
王泰俐 (2003) 網路政治傳播的互動效果。《選舉研究》，10卷2期，P93-134。
教育部國民及學前教育署 (2016)。十二年國民基本教育課程綱要科技領域。
黃國禎、朱蕙君、曾秋蓉、黃國豪、黃繼緯、林農堯 (2007)。具自我調適功能之線上課程問題自動回覆系統。《電子商務學報》，9(3)。


英文文獻
Blei, D. M., Ng, A. Y., and Jordan, M. I. (2003). Latent dirichlet allocation. The Journal of Machine Learning Research, vol. 3, pp. 993–1022.
Carracedo, P., Puertas, R., Marti, L. (2020). Research lines on the impact of the COVID-19 pandemic on business. A text mining analysis. Journal of Business Research. vol. 132, pp. 586-593.
Choy, S. O. and Ng, S. C. (2004). An interactive learning environment for teaching and learning of computer programming. IEEE International Conference on Advanced Learning Technologies, Proceedings., pp. 848-849, 
Corritore, C. L., Wiedenbeck, S. (1991). What do novices learn during program comprehension? International Journal of Human-Computer Interaction, vol.3, Issue.2, pp. 199-222. 
Erra, U., Senatore, S., Minnella, F., Caggianese, G. (2015). Approximate TF–IDF based on topic extraction from massive message stream using the GPU. Information Sciences, vol. 292, pp. 143-161 
GitHub, [online] Available: https://github.com, [Accessed: 05-May-2021] 
Ismail, M. N., Ngah, N. A. and Umar, I. N. (2010). Instructional strategy in the teaching of computer programming: A need assessment analyses. Turkish Online Journal of Educational Technology, vol. 9, no. 2, pp. 125-131.
Jenkins, T. (2002). On the difficulty of learning to program. In Proceedings of the 3rd Annual Conference of the LTSN Centre for Information and Computer Sciences, vol. 4, pp. 53 – 58 
Kang, A., Ren, L., Hua, C., Song, H., Dong, M., Fang, Z., Zhu, M. (2021). Environmental management strategy in response to COVID-19 in China: Based on text mining of government open information. Science of The Total Environment, vol. 769, 145-158.
Kumar, S., Kar, A. K., Ilavarasan, P. V. (2021). Applications of text mining in services management: A systematic literature review. International Journal of Information Management Data Insights, vol. 1, Issue 1, article. 100008 
Matthews, R., Hin, H. S. and Choo, K. A. (2009). Multimedia learning object to build cognitive understanding in learning introductory programming. Proceedings of the 7th international Conference on Advances in Mobile Computing and Multimedia, pp. 396-400.
MDN Web Docs, [online] Available: https://developer.mozilla.org/, [Accessed: 07-May-2021]. 
Ortiz, O., Alcover, P. M., Sánchez, F., Pastor, J. Á. and Herrero, R. (2015). M-Learning Tools: The Development of Programming Skills in Engineering Degrees. IEEE Revista Iberoamericana de Tecnologias del Aprendizaje, vol. 10, no. 3, pp. 86-91. 
Package ‘LDAvis’, [online] Available: https://cran.r-project.org/web/packages/LDAvis
/LDAvis.pdf, [Accessed: 22-July-2021].
Pedregosa, F., Varoquaux, G., Gramfort, A., et al. (2011). Scikit-learn: machine learning in Python. Journal of Machine Learning Research, vol. 12, pp. 2825-2830 
Rau, P. L. P., Chen, S. H., Chin, Y. T. (2004). Developing web annotation tools for learners and instructors. Interacting with Computers, vol. 16, Issue 2, pp. 163-181. 
Shen, X. and Qin, R. (2021). Searching and learning english translation long text information based on heterogeneous multiprocessors and data mining. Microprocessors and Microsystems, vol. 82, article. 103895. 
Song, D., Hong, H., Oh, E. Y. (2021). Applying computational analysis of novice learners' computer programming patterns to reveal self-regulated learning, computational thinking, and learning performance. Computers in Human Behavior, vol. 120, article. 106746.
Stack Overflow, [online] Available: https://insights.stackoverflow.com/survey, [Accessed: 05-May-2021].
Tan, P. H., Ting, C. Y. and Ling, S. W. (2009). Learning difficulties in programming courses: undergraduates' perspective and perception. In Computer Technology and Development. International Conference on ICCTD'09, vol. 1, pp. 42-46.
Tang, Z., Li, W., Li, Y., Zhao. W., Li, S. (2020). Several alternative term weighting methods for text representation and classification. Knowledge-Based Systems, vol.207. article. 106399.
Thomas, P. and Paine, C. (2001). How students learn to program: observations of practical tasks completed. Proceedings IEEE International Conference on Advanced Learning Technologies, pp. 170-173. 
Tibshirani, R., Walther, G. and Hastie, T. (2001). Estimating the number of clusters in a data set via the gap statistic. Journal of the Royal Statistical Society: Series B, vol. 63, pp. 411-423.
Vue.js, [online] Available: https://vuejs.org, [Accessed: 28-Mar-2021].
Xie, X., Fu, Y., Jin, H., Zhao, Y., Cao, W. (2020). A novel text mining approach for scholar information extraction from web content in Chinese. Future Generation Computer Systems, vol. 111, pp. 859-872.
Zhang, T., Sahinidis, N. V., Rosé, C. P., Amaran, S., Shuang, B. (2019). Forty years of Computers and Chemical Engineering: Analysis of the field via text mining techniques. Computers & Chemical Engineering, vol. 129, article. 06511.