本研究的目的是希望藉由分析社群媒體資料對有憂鬱傾向者有更多的了解，尤其其中可能包含從未接觸過精神醫療服務的使用者，並希望運用機器學習技術增進精神醫療臨床實務效益。
本研究的方法是以社群論壇「批踢踢實業坊」(PTT)作為文本資料來源，共92,273筆，並以人工標註出有表達自殺意念的365筆資料。運用機器學習技術建立有憂鬱傾向者的情緒狀態偵測模型(偵測是快樂或悲傷)和自殺意念偵測模型。並以獨立樣本t檢定和卡方檢定分析文本資料，以了解有和無憂鬱傾向者於普遍性、快樂和悲傷情緒狀態，以及表達自殺意念時的文本特性差異(包含發文時間、人稱代名詞使用頻率和發文字數等變項)。
本研究的結果顯示情緒狀態偵測模型預測能力可達AUC(Area Under the Receiver Operating Characteristic Curve) = .889。而自殺意念偵測模型預測能力可達AUC = .964，但AUPRC(Area Under the Precision-Recall Curve) = .315，顯示該模型對於偵測有自殺意念的使用者預測能力較低，但因臺灣尚未有相關研究，本研究的初步探索可供借鏡。而有和無憂鬱傾向者的文本特性差異分析結果顯示於發文時間、人稱代名詞使用頻率和發文字數有明顯差異，並且於快樂情緒狀態時的差異較大，但表達自殺意念時的差異不明顯。本研究結果未來可能運用於早期偵測和心理衡鑑資訊蒐集，以及追蹤治療後情緒變化，以協助臨床決策判斷等。但本研究因受限於難以取得適當且足夠的自殺相關文本資料，相關分析結果待後續研究進一步檢驗和改善。

The purpose of this study was to have a better understanding of people with depressive tendencies by analyzing social media data, especially some data probably from users who never had contact with psychiatric services. Another aim was to use machine learning techniques to improve the efficiency of psychiatric clinical practice.
This study’s data were from the internet forum “PTT” with a total of 92,273 articles. Among these, there were 365 annotated articles expressing suicide ideation. This study used machine learning techniques to build the “Mood state detection model” (detection of happy-sad mood) and the “Suicide ideation detection model” for people with depressive tendencies. And this study analyzed data with variables of the timing of texts, frequency of personal pronoun use, and word counts by independent t-test and chi-square testing to know the difference of textual features between people with and without depressive tendencies in general, happy, sad, and suicide-ideation states.
The results of this study were that the mood state detection model’s AUC(Area Under the Receiver Operating Characteristic Curve) = .889. And the suicide ideation detection model’s AUC = .964, but AUPRC(Area Under the Precision-Recall Curve) = .315. It means this model has low predictability for suicide ideation of people with depressive tendencies. However, there is no related research in Taiwan, the preliminary exploration of this model can be for future reference. The statistical results show there were significant differences of textual features between users with and without depressive tendencies in the timing of texts, frequency of personal pronoun use, and word counts, especially when users were expressing happy feelings, but no significant differences when users were expressing suicide ideation. The results of this study have the potential to be early detection tools, help collect information for psychological assessment, and track emotion trends after medical therapy for enhancing clinical decision-making. However, this study couldn’t get enough representative data about suicide ideation, related results need to be further inspected and improved by follow-up research.

第一章 緒論 1
1-1、研究緣起與動機 1
1-2、研究目的與假設 2
1-3、名詞解釋 5
第二章 文獻探討 15
2-1、應用機器學習技術分析社群媒體資料運用於臨床心理領域 15
2-2、應用機器學習技術以文本資料偵測情緒狀態 16
2-3、應用機器學習技術以文本資料偵測自殺意念 18
2-4、跨國比較有和無憂鬱傾向者文本特性差異 19
第三章 研究方法 23
3-1、文本資料描述和預處理 23
3-2、「情緒狀態偵測模型」和「自殺意念偵測模型」建立步驟 28
3-3、有和無憂鬱傾向者文本特性差異分析 32
第四章 研究結果 35
4-1、情緒狀態偵測模型 35
4-2、有和無憂鬱傾向者的文本特性差異 37
4-3、有和無憂鬱傾向者表達快樂和悲傷情緒狀態時的文本特性差異 41
4-4、自殺意念偵測模型 49
4-5、有和無憂鬱傾向者表達自殺意念時的文本特性差異 51
第五章 討論 55
5-1、情緒狀態偵測模型和相關文本特性分析 55
5-2、自殺意念偵測模型和相關文本特性分析 59
5-3、研究貢獻與未來應用 60
5-4、研究限制與未來研究方向 62
第六章 結論 65
參考文獻 67
附錄 75

“[公告] Prozac板板規4.0版-104/03/31啟用” (2015, March 31). 批踢踢實業坊. https://www.pttweb.cc/bbs/prozac/M.1427795774.A.53C
“[公告] 八卦板板規 (2021, May 11)”. 批踢踢實業坊. https://www.ptt.cc/bbs/Gossiping/M.1620716589.A.F0C.html
“[公告] 板規v1.6” (2009, October 10). 批踢踢實業坊. https://www.ptt.cc/bbs/Sad/M.1255172525.A.F91.html
“[公告] 黑皮版飼養的烏龜2008.10月上路”(2008, September 26). 批踢踢實業坊. https://www.ptt.cc/man/happy/M.1240279431.A.CB5.html
“What is Ptt?” (n.d.). 批踢踢實業坊. Retrieved February 9, 2020, from https://www.ptt.cc/index.html
中央研究院CKIP Lab中文詞知識庫小組(2019). CkipTagger開源中文處理工具. https://github.com/ckiplab/ckiptagger/wiki/Chinese-README
批踢踢 (2019, November 29). In Wikipedia. https://zh.wikipedia.org/w/index.php?title=批踢踢&oldid=57067305
國家衛生研究院 & 衛生福利部國民健康署 (2017). 民國106年國民健康訪問調查. https://www.hpa.gov.tw/Pages/Detail.aspx?nodeid=364&pid=13636
張本聖, 高振傑, 胡淑娥 & 洪志美(譯)(2019). 心理衡鑑大全 (原作者：Groth-Marnat, G. & Wright, A. J.) 雙葉書廊. (原著出版年：2016)
黃金蘭, 林以正, 謝亦泰, & 程威銓. (2012). 中文版「語文探索與字詞計算」詞典之建立. 中華心理學刊, 54(2), 185-201. http://dx.doi.org/10.6129%2FCJP.2012.5402.04
黃敏偉. (2020). 人工智慧(AI)是否可幫助自殺防治？. 自殺防治網通訊, 15(1), 2.
衛生福利部中央健康保險署. (2021). 抗憂鬱藥物使用人數. 政府資料開放平臺. https://data.gov.tw/dataset/146577
衛生福利部心理及口腔健康司. (2020). 自殺死亡及自殺通報統計. https://dep.mohw.gov.tw/domhaoh/cp-4904-8883-107.html
鄭泰安. (2013). 近 20 年台灣焦慮症與憂鬱症盛行率倍增. 當代醫學, (472), 91-94. http://dx.doi.org/10.29941%2FMT.201302_(472).0005
謝佩芸 & 吳佳儀. (2020). 人工智慧暨新興科技於網路自殺防治領域之應用與發展. 自殺防治網通訊, 15(1), 4-6.
Alam, M. (2020). Data normalization in machine learning. Medium. https://towardsdatascience.com/data-normalization-in-machine-learning-395fdec69d02
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (DSM-5®). American Psychiatric Pub.
Bernert, R. A., Hilberg, A. M., Melia, R., Kim, J. P., Shah, N. H., & Abnousi, F. (2020). Artificial intelligence and suicide prevention: a systematic review of machine learning investigations. International journal of environmental research and public health, 17(16), 5929. https://doi.org/10.3390/ijerph17165929
Binali, H., Wu, C., & Potdar, V. (2010, April). Computational approaches for emotion detection in text. In 4th IEEE International Conference on Digital Ecosystems and Technologies, Dubai, United Arab Emirates. https://doi.org/10.1109/DEST.2010.5610650
Cheng, Q., & Lui, C. S. M. (2021). Applying text mining methods to suicide research. Suicide and Life‐Threatening Behavior, 51(1), 137-147. https://doi.org/10.1111/sltb.12680
Cheng, Q., Li, T. M., Kwok, C. L., Zhu, T., & Yip, P. S. (2017). Assessing suicide risk and emotional distress in Chinese social media: A text mining and machine learning study. Journal of medical internet research, 19(7):e243. https://doi.org/10.2196/jmir.7276
Conway, M., & O’Connor, D. (2016). Social media, big data, and mental health: current advances and ethical implications. Current opinion in psychology, 9, 77-82. https://doi.org/10.1016/j.copsyc.2016.01.004
DataReportal (2020). DIGITAL 2020: TAIWAN. https://datareportal.com/reports/digital-2020-taiwan
Davis, J., & Goadrich, M. (2006, June). The relationship between Precision-Recall and ROC curves. Proceedings of the 23rd international conference on Machine learning, Pennsylvania, USA. https://doi.org/10.1145/1143844.1143874
De Choudhury, M., Gamon, M., Counts, S., & Horvitz, E. (2013, June). Predicting depression via social media. Seventh International AAAI Conference on Weblogs and Social Media, 7(1), 128-137. https://ojs.aaai.org/index.php/ICWSM/article/view/14432
Eichstaedt, J. C., Smith, R. J., Merchant, R. M., Ungar, L. H., Crutchley, P., Preoţiuc-Pietro, D., Asch, D. A. & Schwartz, H. A. (2018). Facebook language predicts depression in medical records. Proceedings of the National Academy of Sciences, 115(44), 11203-11208. https://doi.org/10.1073/pnas.1802331115
Ekman, P., & Friesen, W. V. (1971). Constants across cultures in the face and emotion. Journal of Personality and Social Psychology, 17(2), 124–129. https://doi.org/10.1037/h0030377
Evans-Lacko, S., Aguilar-Gaxiola, S., Al-Hamzawi, A., Alonso, J., Benjet, C., Bruffaerts, R., Chiu, W. T., Florescu, S., de Girolamo, G., Gureje, O., Haro, J. M., He, Y., Hu., C., Karam, E. G., Kawakami, N., Lee, S., Lund, C., Kovess-Masfety, V., Levinson, D., ... Thornicroft, G. (2018). Socio-economic variations in the mental health treatment gap for people with anxiety, mood, and substance use disorders: results from the WHO World Mental Health (WMH) surveys. Psychological medicine, 48(9), 1560-1571. https://doi.org/10.1017/S0033291717003336
Geschwind, N., Nicolson, N. A., Peeters, F., van Os, J., Barge-Schaapveld, D., & Wichers, M. (2011). Early improvement in positive rather than negative emotion predicts remission from depression after pharmacotherapy. European Neuropsychopharmacology, 21(3), 241-247. https://doi.org/10.1016/j.euroneuro.2010.11.004
Go, A., Bhayani, R., & Huang, L. (2009). Twitter sentiment classification using distant supervision. CS224N project report, Stanford, 1(12), 2009.
Grus, J. (2015). Data science from scratch: first principles with python. (pp. 142) O'Reilly Media.
Guntuku, S. C., Yaden, D. B., Kern, M. L., Ungar, L. H., & Eichstaedt, J. C. (2017). Detecting depression and mental illness on social media: an integrative review. Current Opinion in Behavioral Sciences, 18, 43-49. https://doi.org/10.1016/j.cobeha.2017.07.005
Guo, Y. (2017). The 7 Steps of Machine Learning. Medium. https://towardsdatascience.com/the-7-steps-of-machine-learning-2877d7e5548e
Henderson, C., Evans-Lacko, S., & Thornicroft, G. (2013). Mental illness stigma, help seeking, and public health programs. American journal of public health, 103(5), 777-780. https://dx.doi.org/10.2105%2FAJPH.2012.301056
Huang, J., & Ling, C. X. (2005). Using AUC and accuracy in evaluating learning algorithms. IEEE Transactions on knowledge and Data Engineering, 17(3), 299-310. https://doi.ieeecomputersociety.org/10.1109/TKDE.2005.50
Inkster, B., Stillwell, D., Kosinski, M., & Jones, P. (2016). A decade into Facebook: where is psychiatry in the digital age?. The Lancet Psychiatry, 3(11), 1087-1090. https://doi.org/10.1016/s2215-0366(16)30041-4
Ji, S., Pan, S., Li, X., Cambria, E., Long, G., & Huang, Z. (2020). Suicidal ideation detection: A review of machine learning methods and applications. IEEE Transactions on Computational Social Systems, 8(1), 214-226. https://doi.org/10.1109/TCSS.2020.3021467
Kemp, S. (2020). DIGITAL 2020: 3.8 BILLION PEOPLE USE SOCIAL MEDIA. We Are Social. https://wearesocial.com/blog/2020/01/digital-2020-3-8-billion-people-use-social-media
Kring, A. M., Davison, G. C., Neale, J. M., & Johnson, S. L. (2016). Abnormal psychology 13E. John Wiley & Sons Inc.
Le Glaz, A., Haralambous, Y., Kim-Dufor, D. H., Lenca, P., Billot, R., Ryan, T. C., Marsh, J., DeVylder, J., Walter, M., Berrouiguet, S. & Lemey, C. (2021). Machine learning and natural language processing in mental health: Systematic review. Journal of Medical Internet Research, 23(5), e15708. https://doi.org/10.2196/15708
Lin, M., Lucas Jr, H. C., & Shmueli, G. (2013). Research commentary—too big to fail: large samples and the p-value problem. Information Systems Research, 24(4), 906-917. https://www.jstor.org/stable/24700283
Locke, S. D., & Gilbert, B. O. (1995). Method of psychological assessment, self-disclosure, and experiential differences: A study of computer, questionnaire, and interview assessment formats. Journal of Social Behavior and Personality, 10(1), 255.
McKinney, W. (2010). Data structures for statistical computing in python. Proceedings of the 9th Python in Science Conference. Texas, USA. https://doi.org/10.25080/MAJORA-92BF1922-00A
Nock, M., Borges, G., Bromet, E., Alonso, J., Angermeyer, M., Beautrais, A., Bruffaerts, R., Chiu, W. T., de Girolamo, G., Gluzman, S., de Graaf, R., Gureje, O., Haro, J. M., Huang, Y., Karam, E., Kessler, R. C., Lepine, J. P., Levinson, D., Medina-Mora, M. E., ... Williams, D. (2008). Cross-national prevalence and risk factors for suicidal ideation, plans and attempts. British Journal of Psychiatry, 192(2), 98-105. https://doi.org/10.1192/bjp.bp.107.040113
Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). Rethinking rumination. Perspectives on psychological science, 3(5), 400-424. https://doi.org/10.1111%2Fj.1745-6924.2008.00088.x
Ozenne, B., Subtil, F., & Maucort-Boulch, D. (2015). The precision–recall curve overcame the optimism of the receiver operating characteristic curve in rare diseases. Journal of clinical epidemiology, 68(8), 855-859. https://doi.org/10.1016/j.jclinepi.2015.02.010
Padrez, K. A., Ungar, L., Schwartz, H. A., Smith, R. J., Hill, S., Antanavicius, T., Brown, D. M., Crutchley, P., Asch, D. A., & Merchant, R. M. (2016). Linking social media and medical record data: a study of adults presenting to an academic, urban emergency department. BMJ quality & safety, 25(6), 414-423. http://dx.doi.org/10.1136/bmjqs-2015-004489
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel. M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., & Duchesnay, E. (2011). Scikit-learn: Machine learning in Python. the Journal of machine Learning research, 12, 2825-2830. https://www.jmlr.org/papers/v12/pedregosa11a.html
Purse, M. (2020, March 25). What Is Suicidal Ideation? Verywell Mind. https://www.verywellmind.com/suicidal-ideation-380609
Python. (2021, Octobe 26). In Wikipedia. https://zh.wikipedia.org/wiki/Python
Raschka, S., & Mirjalili, V. (2019). Python Machine Learning: Machine Learning and Deep Learning with Python, scikit-learn, and TensorFlow 2(3rd edition). Packt Publishing Ltd.
Resnik, P., De Choudhury, M., Schafer, K. M., & Coppersmith, G. (2021). Bibliometric Studies and the Discipline of Social Media Mental Health Research. Comment on “Machine Learning for Mental Health in Social Media: Bibliometric Study”. Journal of Medical Internet Research, 23(6), e28990. https://doi.org/10.2196/28990
Schwartz, H. A., Eichstaedt, J., Kern, M., Park, G., Sap, M., Stillwell, D., Kosinski, M., & Ungar, L. (2014, June). Towards assessing changes in degree of depression through facebook. In Proceedings of the workshop on computational linguistics and clinical psychology: from linguistic signal to clinical reality, Maryland, USA. https://aclanthology.org/W14-3214
Suicide Prevention Resource Center. (n.d.). Topics and Terms. Retrieved December 3, 2021 from https://www.sprc.org/about-suicide/topics-terms
The World Bank. (n.d.). Taiwan, China | Data. Retrieved December 3, 2021 from https://data.worldbank.org/country/TW
Wongkoblap, A., Vadillo, M. A., & Curcin, V. (2017). Researching mental health disorders in the era of social media: systematic review. Journal of medical Internet research, 19(6), e228. https://doi.org/10.2196/jmir.7215
World Health Organization. (n.d.). Suicide. Retrieved December 3, 2021 from http://www.emro.who.int/health-topics/suicide/feed/atom.html
Xu, H., Yang, W., & Wang, J. (2015). Hierarchical emotion classification and emotion component analysis on Chinese micro-blog posts. Expert systems with applications, 42(22), 8745-8752. https://doi.org/10.1016/j.eswa.2015.07.028
Yuan, Z. & Purver, M. (2015) Predicting Emotion Labels for Chinese Microblog Texts. In Gaber, M., Cocea, M., Wiratunga, N. & Goker, A. (eds), Advances in Social Media Analysis (pp 129-149). Springer, Cham. https://doi.org/10.1007/978-3-319-18458-6_7