Fine-Grained emotion classification in reddit mental health posts using LLMs

Vista sencilla de ítem
dc.contributor.advisorGalpin, Ixent
dc.creatorOrjuela Llanos, Valery Pamela
dc.date.accessioned2025-06-16T22:15:32Z
dc.date.available2025-06-16T22:15:32Z
dc.date.created2025-06-16
dc.description.abstractEste estudio evalúa el discurso sobre salud mental en Reddit a través del análisis de emociones basado en IA. A partir de un conjunto de datos exhaustivo que incluye 7899 publicaciones y 54755 comentarios sobre 10 afecciones relacionadas con la salud mental, este estudio combina técnicas de PNL como BERTopic y el análisis léxico, con el uso de GPT-4o-mini para clasificar las emociones. Los resultados indican claras diferencias emocionales entre publicaciones y comentarios: mientras que los títulos suelen expresar urgencia (p. ej., "ayuda"), los comentarios muestran empatía (p. ej., "amor"). GPT-4o-mini coincidió con las anotaciones de emociones humanas con una precisión del 52,6 %, obteniendo un mejor rendimiento con las emociones positivas (58 %). El análisis temático reveló que los diagnósticos clínicos predominaban en las publicaciones, pero los comentarios contenían diálogos de apoyo. Este trabajo ilustró el potencial de los LLM como herramienta para el monitoreo de la salud mental y el bienestar, a la vez que indica posibles inconvenientes con respecto a la clasificación ambigua de emociones. El estudio también ofrece información práctica para el desarrollo de herramientas de IA en el ámbito de la salud mental y describe direcciones para el trabajo futuro.spa
dc.description.abstractenglishThis study evaluates the discourse surrounding mental health on Reddit through the lens of AI-based emotion analysis. Drawing from a comprehensive dataset comprising 7,899 posts and 54,755 comments on 10 mental health related conditions, this study combines NLP techniques such as BERTopic and lexical analysis, and the usage of GPT-4o-mini to classify emotions. The results point to clear emotional differences between posts and comments: while titles often express urgency (e.g., ’help’), comments show empathy (e.g., ’love’). GPT-4o-mini matched human emotion annotations with an accuracy of 52.6%, performing better with positive emotions (58%). Thematic analysis revealed that clinical diagnoses were dominant in the posts, but the comments contained supportive dialogue. This work illustrated that LLMs have potential as an avenue for monitoring mental health and well-being, while also indicating potential drawbacks with respect to ambiguous emotion classification. The study also offers practical information for the development of AI tools in the mental health space and outlines directions for future work.spa
dc.format.extent16 páginasspa
dc.format.mimetypeapplication/pdfspa
dc.identifier.urihttps://hdl.handle.net/20.500.12010/36877
dc.language.isoengspa
dc.relation.referencesAl-Ansari, Ahmed, J.: From sentiment analysis to choreography of emotions: social media analysis for improved customer relationship management (CRM) in the Omani telecom sector. Ph.D. thesis, University of Strathclyde (2019)spa
dc.relation.referencesAtari, M., Mehl, M.R., Graham, J., Doris, J.M., Schwarz, N., Davani, A.M., Om- rani, A., Kennedy, B., Gonzalez, E., Jafarzadeh, N., Hussain, A., Mirinjian, A., Madden, A., Bhatia, R., Burch, A., Harlan, A., Sbarra, D.A., Raison, C.L., Mose- ley, S.A., Polsinelli, A.J., Dehghani, M.: The paucity of morality in everyday talk. Scientific Reports 13, 5967 (4 2023). https://doi.org/10.1038/s41598-023-3 2711-4spa
dc.relation.referencesBarrett, L.F.: Are Emotions Natural Kinds? Perspectives on Psychological Science 1, 28–58 (3 2006). https://doi.org/10.1111/j.1745-6916.2006.00003.xspa
dc.relation.referencesCenter, P.: Teen Internet Access Demographics, Pew Internet and American Life Project (2012)spa
dc.relation.referencesCha, J., Kim, S., Park, E.: A lexicon-based approach to examine depression de- tection in social media: the case of Twitter and university community. Humanities and Social Sciences Communications 9, 325 (9 2022). https://doi.org/10.105 7/s41599-022-01313-2spa
dc.relation.referencesCoppersmith, G., Dredze, M., Harman, C., Hollingshead, K.: From ADHD to SAD: Analyzing the Language of Mental Health on Twitter through Self-Reported Diag- noses.In:Proceedingsofthe2ndWorkshoponComputationalLinguisticsandClin- ical Psychology: From Linguistic Signal to Clinical Reality. pp. 1–10. Association for Computational Linguistics (2015). https://doi.org/10.3115/v1/W15-1201spa
dc.relation.referencesDante, Durante, R., Sobbrio, F., Zejcirovic, D.D.: Lost in the net? broadband internet and youth mental health. JSTOR (2022)spa
dc.relation.referencesDemszky,D.,Movshovitz-Attias,D.,Ko,J.,Cowen,A.,Nemade,G.,Ravi,S.:GoE- motions: A Dataset of Fine-Grained Emotions. arXiv preprint arXiv:2005.00547 (5 2020)spa
dc.relation.referencesEgger, R., Yu, J.: A Topic Modeling Comparison Between LDA, NMF, Top2Vec, and BERTopic to Demystify Twitter Posts. Frontiers in Sociology 7 (5 2022). https://doi.org/10.3389/fsoc.2022.886498spa
dc.relation.referencesEichstaedt, J.C., Smith, R.J., Merchant, R.M., Ungar, L.H., Crutchley, P., Preoţiuc-Pietro, D., Asch, D.A., Schwartz, H.A.: Facebook language predicts de- pression in medical records. Proceedings of the National Academy of Sciences 115, 11203–11208 (10 2018). https://doi.org/10.1073/pnas.1802331115spa
dc.relation.referencesEuronews: ChatGPT a year on: 3 ways the AI chatbot has completely changed the world in 12 months (11 2023), https://www.euronews.com/next/2023/11/30/c hatgpt-a-year-on-3-ways-the-ai-chatbot-has-completely-changed-the-w orld-in-12-monthsspa
dc.relation.referencesGrootendorst, M.: BERTopic: Neural topic modeling with a class-based TF-IDF procedure. arXiv preprint arXiv:2203.05794 (3 2022)spa
dc.relation.referencesGWI: Social: GWI’s Flagship Report on the Latest Trends in Social Media. Tech. rep., GWI (2023)spa
dc.relation.referencesHaythornthwaite, C., Kumar, P., Gruzd, A., Gilbert, S., del Valle, M.E., Paulin, D.: Learning in the wild: coding for learning and practice on Reddit. Learning, Media and Technology 43, 219–235 (7 2018). https://doi.org/10.1080/174398 84.2018.1498356spa
dc.relation.referencesJose, R., Matero, M., Sherman, G., Curtis, B., Giorgi, S., Schwartz, H.A., Un- gar, L.H.: Using Facebook language to predict and describe excessive alcohol use. Alcoholism: Clinical and Experimental Research 46, 836–847 (5 2022). https: //doi.org/10.1111/acer.14807spa
dc.relation.referencesKheiri, K., Karimi, H.: SentimentGPT: Exploiting GPT for Advanced Senti- ment Analysis and its Departure from Current Machine Learning. arXiv preprint arXiv:2307.10234 (7 2023)spa
dc.relation.referencesKim, J., Lee, J.E.R.: The Facebook Paths to Happiness: Effects of the Num- ber of Facebook Friends and Self-Presentation on Subjective Well-Being. Cy- berpsychology, Behavior, and Social Networking 14, 359–364 (6 2011). https: //doi.org/10.1089/cyber.2010.0374spa
dc.relation.referencesLee, J., Kim, C.: A Structure of basic emotions: A review of basic emotion theories usingan emotionallyfine-tunedlanguagemodel,vol.45.Proceedings oftheAnnual Meeting of the Cognitive Science Society (2023)spa
dc.relation.referencesOcal, A.: Framing, Emotions, Salience: The Future Of AI As Seen By Redditors. Ph.D. thesis, Syracuse University (2023)spa
dc.relation.referencesOcal, A.: Perceptions of the Future of Artificial Intelligence on Social Media: A Topic Modeling and Sentiment Analysis Approach. IEEE Access 12, 182386– 182409 (2024). https://doi.org/10.1109/ACCESS.2024.3510526spa
dc.relation.referencesOdilboyevich, J.X.: Psychological Description of Teenage Outlaws in Interpersonal Relationships. Journal of Preschool Education and Psychology 1, 20–22 (2024)spa
dc.relation.referencesOmuya, E., Okeyo, G., Kimwele, M.: Sentiment Analysis on Social Media using Machine Learning Approach (11 2021). https://doi.org/10.22541/au.16362014 3.37655829/v1spa
dc.relation.referencesOrganization., W.H.: Mental disorders. (6 2022), https://www.who.int/news-r oom/fact-sheets/detail/mental-disordersspa
dc.relation.referencesParis, C., Christensen, H., Batterham, P., O’Dea, B.: Exploring Emotions in Social Media.In:2015IEEEConferenceonCollaborationandInternetComputing(CIC). pp. 54–61. IEEE (10 2015). https://doi.org/10.1109/CIC.2015.43spa
dc.relation.referencesPirina, I., cagri coltekin: Identifying Depression on Reddit: The Effect of Training Data. In: Proceedings of the 2018 EMNLP Workshop SMM4H: The 3rd Social Media Mining for Health Applications Workshop and Shared Task. pp. 9–12. As- sociation for Computational Linguistics (2018). https://doi.org/10.18653/v1/ W18-5903spa
dc.relation.referencesPlutchik, R., Williams, M., Jerrett, I., Karasu, T.B., Kane, C.: Emotions, person- ality and life stresses in asthma. Journal of Psychosomatic Research 22, 425–431 (1 1978). https://doi.org/10.1016/0022-3999(78)90065-Xspa
dc.relation.referencesProferes, N., Jones, N., Gilbert, S., Fiesler, C., Zimmer, M.: Studying Reddit: A Systematic Overview of Disciplines, Approaches, Methods, and Ethics. Social Media + Society 7 (4 2021). https://doi.org/10.1177/20563051211019004spa
dc.relation.referencesPénard, T., Poussing, N., Suire, R.: Does the Internet make people happier? The Journal of Socio-Economics 46, 105–116 (10 2013). https://doi.org/10.1016/j. socec.2013.08.004spa
dc.relation.referencesRathje, S., Mirea, D.M., Sucholutsky, I., Marjieh, R., Robertson, C.E., Bavel, J.J.V.: GPT is an effective tool for multilingual psychological text analysis. Pro- ceedings of the National Academy of Sciences 121 (8 2024). https://doi.org/10 .1073/pnas.2308950121spa
dc.relation.referencesValenzuela, S., Park, N., Kee, K.F.: Is There Social Capital in a Social Network Site?: Facebook Use and College Students’ Life Satisfaction, Trust, and Partic- ipation. Journal of Computer-Mediated Communication 14, 875–901 (7 2009). https://doi.org/10.1111/j.1083-6101.2009.01474.xspa
dc.relation.referencesWankhade, M., Rao, A.C.S., Kulkarni, C.: A survey on sentiment analysis meth- ods, applications, and challenges. Artificial Intelligence Review 55, 5731–5780 (10 2022). https://doi.org/10.1007/s10462-022-10144-1spa
dc.relation.referencesWei, J., Bosma, M., Zhao, V.Y., Guu, K., Yu, A.W., Lester, B., Du, N., Dai, A.M., Le, Q.V.: Finetuned Language Models Are Zero-Shot Learners. arXiv preprint arXiv:2109.01652 (9 2021)spa
dc.relation.referencesWen, R., Crowe, S.E., Gupta, K., Li, X., Billinghurst, M., Hoermann, S., Allan, D., Nassani, A., Piumsomboon, T.: Large Language Models for Automatic Detection of Sensitive Topics. arXiv preprint arXiv:2409.00940 (9 2024)spa
dc.relation.referencesWillcox, G.: The Feeling Wheel. Transactional Analysis Journal 12, 274–276 (10 1982). https://doi.org/10.1177/036215378201200411spa
dc.relation.referencesZad,S.,Heidari,M.,Jones,J.H.J.,Uzuner,O.:EmotionDetectionofTextualData: An Interdisciplinary Survey. In: 2021 IEEE World AI IoT Congress (AIIoT). pp. 0255–0261.IEEE(52021).https://doi.org/10.1109/AIIoT52608.2021.9454192spa
dc.relation.referencesZhan, T., Shi, C., Shi, Y., Li, H., Lin, Y.: Optimization Techniques for Sentiment Analysis Based on LLM (GPT-3) (5 2024)spa
dc.subjectModelos de lenguaje grandes (LLM)
dc.subjectReddit
dc.subjectSalud mental
dc.subjectProcesamiento del lenguaje natural (PLN).
dc.subjectAnálisis de emocionesspa
dc.subject.keywordLarge language models (LLMs)
dc.subject.keywordReddit
dc.subject.keywordMental health
dc.subject.keywordNatural language processing (NLP).
dc.subject.keywordEmotion analysisspa
dc.subject.lembSalud mental - Aspectos sociales - Redes sociales.
dc.subject.lembInteligencia artificial - Aplicaciones en medicina
dc.subject.lembProcesamiento del lenguaje natural (Informática).
dc.titleFine-Grained emotion classification in reddit mental health posts using LLMsspa
dc.type.coarhttp://purl.org/coar/resource_type/c_bdccspa

Archivos

Bloque original

Mostrando 1 - 1 de 1
Miniatura
Nombre:
TesisMaestria_ValeryOrjuela.pdf
Tamaño:
6 MB
Formato:
Adobe Portable Document Format
Descripción:
Documento reservado
Descargar

Bloque de licencias

Mostrando 1 - 2 de 2
Miniatura
Nombre:
license.txt
Tamaño:
2.87 KB
Formato:
Item-specific license agreed upon to submission
Descripción:
Descargar
Miniatura
Nombre:
FOR-EFE-GDB-008_AUTORIZACION_DE_PUBLICACION_DE_TESIS_O_TRABAJO_DE_GRADO_DE_FORMA_CONFIDENCIAL_Valery_IG.pdf
Tamaño:
98.94 KB
Formato:
Adobe Portable Document Format
Descripción:
Carta de autorización
Descargar

Colecciones

Maestría en Ingeniería y Analítica de Datos

Carrera 4 # 22-61 Teléfono: (+57) 601 242 7030 - 018000111022 Fax: (+57) 601 561 2107 Bogotá D.C., Colombia

Fundación Universitaria de Bogotá Jorge Tadeo Lozano | Vigilada Mineducación

Institución de educación superior privada, de utilidad común, sin ánimo de lucro y su carácter académico es el de Universidad

Reconocimiento personería jurídica: Resolución 2613 del 14 de agosto de 1959 Minjusticia.

Institución de Educación Superior sujeta a inspección y vigilancia por el Ministerio de Educación Nacional.

Software DSpace copyright © 2002-2026 LYRASIS