臺灣博碩士論文加值系統

非玩家角色作為獨立的角色，提供了多樣的功能來協助遊戲環境的建構，然而大多時候非玩家角色在遊戲中的行為是固定且可預測的，對於玩家的行動給出機械化的反應。為改善此情形並帶給遊戲更加多元的玩法，情感模擬即是其中一個被提出的方法，透過模擬非玩家角色的情感變化來達成行為多樣化的目的。過去的研究中有多種模型被提出，但每項研究中作為模型基礎所採用的心理學模型不盡相同，在模型建構的細節上也有所分別，在這樣的情況下，本研究提出一種以機器學習方式進行建模的方法，期望在達成一定效果的同時，也降低模型建構上的繁雜性。
本研究使用遊戲中最常用的情感表達方式之一，即遊戲角色間的對白作為資料集來源，並以遊戲《巫師3》中一位遊戲中的主要非玩家角色為研究對象，分析其對白中的情感資訊來建立資料集，並以隨機森林訓練情感模擬模型。本研究以實例闡述了使用角色對白來建立資料集的優點與缺點，並從資料分布與模型訓練的情形說明了為何隨機森林適用於此項任務。實驗結果發現，隨機森林能夠很好的根據資料集進行建模，但資料集的建立需要使用更加精確的方式或是加入更多特徵，才能讓模型展現研究對象的特色。

Serving as independent characters in games, non-player characters (NPCs) provide various functions to assist in constructing the game environment. However, in most cases, the behavior of NPCs is fixed and predictable, resulting in robotic responses to player actions. To improve NPC performance and bring diversity to gameplay, simulating the emotion of NPCs is one of the proposed methods. Previous researches have introduced multiple models, however, there are variations in the psychological models used as the foundation for each study, as well as differences in the details of model construction. As a result, this study proposes a machine learning-based modeling method with the aim of achieving a similar effect to a certain degree while reducing the complexity of the model construction process. This study utilizes dialogues between game characters as the source of data, which is one of the most commonly used forms of emotional expression in games. A major NPC in the game “The Witcher 3” is selected as the research subject. By analyzing its dialogues, the emotional information contained in the dialogues constructs the dataset, and random forest is applied to train the emotion simulation model. We explained the advantages and limitations of using character dialogues to construct a dataset and the suitability of random forest for this task through examples. The experimental results reveal that random forest is capable of modeling, but the construction of the dataset requires more precise methods or the inclusion of additional features to capture the unique characteristics of the research subject NPC.

摘要 i
Abstract ii
圖目錄 v
表目錄 vi
一、緒論 1
1.1 研究動機 1
1.2 研究背景 2
1.2.1 NPC的功能 2
1.2.2 NPC與電子遊戲種類 3
1.2.3遊戲中的情感 6
1.3 研究目的 7
1.4 研究重要性 8
二、 文獻探討 9
2.1 情感模擬 9
2.1.1 情感理論 9
2.1.1 情感模擬之相關研究 10
2.2 文本情感分析 11
2.3 監督式學習 13
三、研究方法 15
3.1 研究架構 15
3.2 研究工具 17
3.3 實驗資料集 18
3.4 研究流程 19
3.4.1 資料集製作 20
3.4.2 模型訓練 22
3.4.3 結果評估 23
四、研究結果 24
4.1 情感分類 24
4.2 情感強度 27
4.3 模型效果 29
4.4 實驗結果分析與討論 32
五、結論與建議 34
5.1 研究發現 34
5.2 研究結論 35
5.3 研究限制和未來方向 35
參考文獻 37
附錄1: 資料集全 39

[1] H. Warpefelt, H. Verhagen. “Towards an updated typology of non-player character roles,” In Game and Entertainment Technologies. IADIS Press, Arhus, Denmark, 2015.
[2] L. Grace. “Game type and game genre,” 2005
[3] Baumeister, R.F.; Vohs, K.D.; DeWall, C.N.; Zhang, L. “How Emotion Shapes Behavior: Feedback, Anticipation, and Reflection, Rather Than Direct Causation.” Personal. Soc. Psychol. Rev. vol.11, pp. 167–203, 2007
[4] Togelius J., Yannakakis G.N., Karakovskiy S., Shaker N. “Assessing Believability,” Believable Bots. Springer, Berlin, Heidelberg, 2013.
[5] J. Bates, “The role of emotions in believable agents,” Communications of the ACM, vol. 37, no. 7, pp. 122–125, 1994.
[6] Lankoski, P, Björk, S. “Gameplay design patterns for believable non-player characters,” Situated Play, Proceedings of DiGRA 2007 Conference. Tokyo, Japan, 2007
[7] H. Kersjes and P. Spronck, "Modeling believable game characters," 2016 IEEE Conference on Computational Intelligence and Games (CIG), Santorini, Greece, 2016, pp. 1-8, doi: 10.1109/CIG.2016.7860412.
[8] J. Broekens, E. Hudlicka,R. Bidarra. “Emotional Appraisal Engines for Games.” Emotion in Games. Socio-Affective Computing, vol 4. Springer, Cham, 2016.
[9] A. Ortony, G. Clore, and A. Collins, The Cognitive Structure of Emotions. Cambridge, U.K.: Cambridge Univ. Press, 1988
[10] Plutchik Robert. Emotion: Theory, research, and experience. vol. 1: Theories of emotion, 1980.
[11] A. Mehrabian, ‘Pleasure-arousal-dominance: A general framework for describing and measuring individual differences in Temperament’, Current Psychology, vol. 14, no. 4, pp. 261–292, Dec. 1996,
[12] Soujanya Poria, Erik Cambria, Rajiv Bajpai, and Amir Hussain. 2017. A review of affective computing. Inf. Fusion 37, C (September 2017), 98–125.
[13] M. Ochs, N. Sabouret and V. Corruble, "Simulation of the Dynamics of Nonplayer Characters' Emotions and Social Relations in Games," IEEE Transactions on Computational Intelligence and AI in Games, vol. 1, no. 4, 2009, pp. 281-297.

[14] H. J. Eysenck, The Biological Basis of Personality, C. C. Thomas, Ed. London, U.K.: Springfield, 1986.
[15] A. Egges, S. Kshirsagar, N. Thalmann, “A model for personality and emotion simulation,” in 7th International Conference on Knowledge-Based and Intelligent Information and Engineering Systems. Springer, 2003, pp. 453-461.
[16] R. R. McCrae and O. P. John, ‘An Introduction to the Five-Factor Model and Its Applications’, Journal of Personality, vol. 60, no. 2, Art. no. 2, Jun. 1992
[17] S. Belle, C. Gittens and T. C. Nicholas Graham, "A Framework for Creating Non-Player Characters That Make Psychologically-Driven Decisions," 2022 IEEE International Conference on Consumer Electronics (ICCE), Las Vegas, NV, USA, 2022, pp. 1-7
[18] F. Garavaglia, R. A. Nobre, L. A. Ripamonti, D. Maggiorini and D. Gadia, "Moody5: Personality-biased agents to enhance interactive storytelling in video games," 2022 IEEE Conference on Games (CoG), Beijing, China, 2022, pp. 175-182
[19] Q. He, "Recent Works for Sentiment Analysis using Machine Learning and Lexicon Based Approaches," 2022 5th International Conference on Advanced Electronic Materials, Computers and Software Engineering (AEMCSE), Wuhan, China, 2022, pp. 422-426,
[20] S. V. Pandey and A. V. Deorankar, "A Study of Sentiment Analysis Task and It's Challenges," 2019 IEEE International Conference on Electrical, Computer and Communication Technologies (ICECCT), Coimbatore, India, 2019, pp. 1-5,
[21] Shotgunnova (2016), The Witcher 3: Wild Hunt – Game Script https://gamefaqs.gamespot.com/pc/699808-the-witcher-3-wild-hunt/faqs/73353
[22] Jochen Hartmann (2022), "Emotion English DistilRoBERTa-base". https://huggingface.co/j-hartmann/emotion-english-distilroberta-base/
[23] Hutto, C., & Gilbert, E. (2014). VADER: A Parsimonious Rule-Based Model for Sentiment Analysis of Social Media Text. Proceedings of the International AAAI Conference on Web and Social Media, 8(1), 216-225.