近年來，人工智慧是一個很熱門的領域，本研究將自然語言處理的技術應用於教育領域，針對「數學文字題」設計出一套「教學輔助對話系統」，該系統著重於解決低成就學生的學習困難，專注於學生的語意理解能力，透過使用者輸入題目到系統，產生互動式的對話或利用代換相似題型的方式輔助，藉以引導學生理解問題，再根據各類型題目提供不同的教學策略，找出學生的學習困難，並且提升低成就學生的學習成效與自我效能感。
本研究提出的系統架構分為三個模組：「題型模式分類模組」、「自然語言理解分析模組」、「對話管理模組」；「題型模式分類模組」是對題目的模式（Pattern）進行分類，我們利用統計準則式方法（Statistical Principle-Based Approach, SPBA），提出一套「模式比對演算法」去做分類，「自然語言理解分析模組」運用語意規則去抽取題目中的重要資訊建構本體論，「對話管理模組」則利用對話腳本產生相對應的教學策略，向學生進行提問與對話。數學文字題的解題在認知心理學中是經過一連串的解題歷程，本論文基於這樣的角度，區分出學生在解題上的各種錯誤類型，並且分別設計解題策略，再藉由與教師訪談結果與問卷調查進行驗證。學生受測結果的對話紀錄顯示，我們的系統是可以有效幫助低成就學生學習，並找出學生的錯誤類型。

In recent years, artificial intelligence is a very popular field. This research applies the technology of natural language processing(NLP) to the field of education. We designs "An Assisted Learning Dialogue System For Math Word Problem", which focuses on solving the learning difficulties of low-achieving students, and students’ semantic comprehension ability. Through users input math word problems into the system, we generate interactive dialogues or use similar problems to guide students to understand the math word problems. We provide different teaching strategies according to various classes of problems, identify students’ learning difficulties, and improve low achievement Students’ learning effectiveness and self-efficacy.
The system architecture proposed in this research is divided into three modules: "Pattern Classification Module", "Natural Language Understanding Analysis Module", and "Dialogue Management module". "Pattern Classification Module" use pattern to classified, we use Statistical Principle-Based Approach (SPBA) to propose a new classification approach called "Pattern Matching Algorithm". "Natural Language Understanding Analysis Module" uses semantic rules to extract information to construct the ontology. "Dialogue Management Module" uses the dialogue script to generate corresponding teaching strategies, and ask questions to students. Solving word math problems is a series of problem-solving processes in cognitive psychology. Based on this perspective, this thesis distinguishes various types of errors that students make in problem-solving, and designs problem-solving strategies respectively, and then interviews with teachers and use questionnaire to verify our strategies. According to the conversation records of the students, our system effectively help students learn, and also find out the types of students’ errors.

[1] R. Mayer, “Educational Psychology: A cognitive approach,” Boston: Little, Brown and
Company, 1987.
[2] G. Polya, “How to solve it,” Princeton, NJ: Princeton University Press, 1973.
[3] A. H. Schoenfeld, “Mathematical Problem Solving,” Orlando, FL: Academic Press, 1985.
[4] J. Garofalo and F. Lester, “Metacognition, cognitive monitoring, and mathematical performance,” journal for Research in Mathematics Education, 16, pp. 163–176, 1987.
[5] C. R. Fletcher, “ Understanding and solving arithmetic word problems: A computer simulation,” Behavior Research Methods, Instruments, Computers, vol. 17, no. 5, pp. 565–571,
Sep 1985.
[6] Y. Bakman, “ Robust Understanding of Word Problems with Extraneous Information,”
ArXiv Mathematics e­prints, 2007.
[7] M. J. Hosseini, H. Hajishirzi, O. Etzioni, and N. Kushman, “ Learning to solve arithmetic
word problems with verb categorization,” EMNLP, pp. 523–533, 2014.
[8] J. Elman, “Finding structure in time,” Cognitive Science, vol. 14, no. 2, pp. 179–211,
1990.
[9] S. Hochreiter and J. Schmidhuber, “Long Short Term Memory. Neural Computation,”
Neural computation,9(8):1735–1780, 1997.
[10] H. Sak, A. Senior, and F. Beaufays., “Long short­term memory recurrent neural network
architectures for large scale acoustic modeling,” Proceedings of the Annual Conference
of International Speech Communication Association (INTERSPEECH), 2014.
[11] Y. Wang, X. Liu, and S. Shi, “ Deep neural solver for math word problems,” EMNLP,
pp. 845–854, 2017.
[12] L. Wang, D. Zhang, L. Gao, J. Song, L. Guo, and H. T. Shen, “ Mathdqn: Solving arithmetic word problems via deep reinforcement learning,” AAAI, 2018.
[13] J. Weizenbaum, “Eliza—a computer program for the study of natural language communication between man and machine,” Commun. ACM, vol. 9, pp. 36–45, Jan. 1966.
[14] O. Vinyals and Q. V. Le, “A neural conversational model,” CoRR, vol. abs/1506.05869,
2015.
[15] L. Shang, Z. Lu, and H. Li, “ Neural responding machine for short­text conversation,”
ACL, pp. 1577–1586, 2015.
60
[16] H. Chen, X. Liu, D. Yin, and J. Tang, “A survey on dialogue systems: Recent advances
and new frontiers,” CoRR, vol. abs/1711.01731, 2017.
[17] T.­H. Wen, D. Vandyke, N. Mrkšić, M. Gašić, L. M. Rojas­Barahona, P.­H. Su, S. Ultes,
and S. Young, “A network­based end­to­end trainable task­oriented dialogue system,”
pp. 438–449, Apr. 2017.
[18] T. Gruber, “ A Translation Approach to Portable Ontologies,” Knowledge Acquisition,
pp. 199–220, 1993.
[19] N. Guarino, “Semantic matching: Formal ontological distinctions for information organization, extraction, and integration,” vol. 1299, May 1998.
[20] J. P. Turian, L. Ratinov, and Y. Bengio, “Word representations: A simple and general
method for semi­supervised learning,” ACL 2010, Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, July 11­ 16, 2010, Uppsala, Sweden,
pp. 384–394, 2010.
[21] T. Mikolov, K. Chen, G. Corrado, and J. Dean, “Efficient estimation of word representations in vector space,” CCoRR, vol. abs/1301.3781, 2013.
[22] J. Devlin, M.­W. Chang, K. Lee, and K. Toutanova, “BERT: Pre­training of deep bidirectional transformers for language understanding,” Proceedings of the 2019 Conference of
the North American Chapter of the Association for Computational Linguistics: Human
Language Technologies, 2019.
[23] R. E. Mayer, “Educational psychology : A cognitive approach,” 1987.
[24] Y. Hsieh, Y. Chang, Y. Huang, C. C. S. Yeh, and W. Hsu, “MONPA: multi­ objective
named­entity and part­of­speech annotator for chinese using recurrent neural network,” in
Proceedings of the Eighth International Joint Conference on Natural Language Processing,
IJCNLP 2017, vol. 2, pp. 80–85, 2017.
[25] Z. Lan, M. Chen, S. Goodman, K. Gimpel, P. Sharma, and R. Soricut, “ ALBERT: A Lite
BERT for Self­supervised Learning of Language Representations,” ICLR, 2020.
[26] M. Cardelle­Elawar, “ Effects of teaching metacognitive skills to students with low mathematics ability,” 1992.
[27] 秦麗花, “數學閱讀指導的理論與實務,” 台北市: 洪葉文化, 2007.