Detecting Emotions in Chat Messages with AI: A Real-Time REST API Built on AWS Bedrock

Nidhi Singh

doi:10.55524/ijirem.2026.13.2.24

Abstract

Most chat platforms today carry an enormous amount of emotional content that nobody is actually tracking. People send billions of messages a day across apps like WhatsApp, Slack and Teams, and a lot of those messages are angry, scared, frustrated, or excited in ways that the platform itself never picks up on. This project tries to do something about that. We built a small REST API that takes a chat message and tells you what emotion is in it. The seven labels we use are Happy, Sad, Angry, Surprised, Fearful, Disgusted, and Neutral. The reply comes back as JSON, and includes the emotion, a confidence number, and a short reason. For the actual classification we use AWS Bedrock so that we don't have to host any models ourselves. Claude 3 Sonnet from Anthropic is the main model. If it fails, we fall back to Amazon's Titan Text Express. The Python code uses FastAPI, runs in Docker on ECS Fargate, and the entire AWS setup is written as code with the CDK in TypeScript. We tested it on a hand-labelled set of 1,000 chat messages and got 88.4% accuracy overall, which is past the 85% target we set at the start. Latency at 50 concurrent users sat at a median of 285 ms, with the 95th percentile at 412 ms. The API also takes the past few messages of a conversation into account when classifying, which helps a lot with sarcasm and one-line follow-ups.

Keywords

AWS Bedrock, Affective Computing, Chat Analysis, Claude Foundation Model, Docker Containers, ECS Fargate, Emotion Detection, FastAPI, Infrastructure as Code, Prompt Engineering

References

B. Pang, L. Lee, and S. Vaithyanathan, “Thumbs up? Sentiment classification using machine learning techniques,” in Proc. Conf. Empirical Methods in Natural Language Processing (EMNLP), Philadelphia, PA, USA, 2002, pp. 79–86. Available from: https://aclanthology.org/W02-1011.pdf
P. D. Turney, “Thumbs up or thumbs down? Semantic orientation applied to unsupervised classification of reviews,” in Proc. Annu. Meeting of the Association for Computational Linguistics (ACL), Philadelphia, PA, USA, 2002, pp. 417–424. Available from: https://aclanthology.org/P02-1053.pdf
S. Baccianella, A. Esuli, and F. Sebastiani, “SentiWordNet 3.0: An enhanced lexical resource for sentiment analysis and opinion mining,” in Proc. Int. Conf. Language Resources and Evaluation (LREC), Valletta, Malta, 2010, pp. 2200–2204. Available from: https://aclanthology.org/L10-1531.pdf
P. Ekman, “An argument for basic emotions,” Cognition and Emotion, vol. 6, no. 3–4, pp. 169–200, May 1992. Available from: https://doi.org/10.1080/02699939208411068
S. M. Mohammad and P. D. Turney, “Crowdsourcing a word-emotion association lexicon,” Computational Intelligence, vol. 29, no. 3, pp. 436–465, Aug. 2013. Available from: https://doi.org/10.1111/j.1467-8640.2012.00460.x
O. Alm, D. Roth, and R. Sproat, “Emotions from text: Machine learning for text-based emotion prediction,” in Proc. Human Language Technology Conf. and Conf. on Empirical Methods in Natural Language Processing (HLT/EMNLP), Vancouver, BC, Canada, 2005, pp. 579–586. Available from: https://aclanthology.org/H05-1073.pdf
S. M. Mohammad, F. Bravo-Marquez, M. Salameh, and S. Kiritchenko, “SemEval-2018 Task 1: Affect in tweets,” in Proc. Int. Workshop on Semantic Evaluation (SemEval), New Orleans, LA, USA, 2018, pp. 1–17. Available from: https://aclanthology.org/S18-1001/
Chatterjee, K. N. Narahari, M. Joshi, and P. Agrawal, “SemEval-2019 Task 3: EmoContext contextual emotion detection in text,” in Proc. Int. Workshop on Semantic Evaluation (SemEval), Minneapolis, MN, USA, 2019, pp. 39–48. Available from: https://aclanthology.org/S19-2005/
M. Abdul-Mageed and L. Ungar, “EmoNet: Fine-grained emotion detection with gated recurrent neural networks,” in Proc. Annu. Meeting of the Association for Computational Linguistics (ACL), Vancouver, BC, Canada, 2017, pp. 718–728. Available from: https://aclanthology.org/P17-1067/
Vaswani et al., “Attention is all you need,” in Advances in Neural Information Processing Systems (NeurIPS), vol. 30, Long Beach, CA, USA, 2017, pp. 5998–6008. Available from: https://papers.nips.cc/paper/7181-attention-is-all-you-need.pdf
J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT), Minneapolis, MN, USA, 2019, pp. 4171–4186. Available from: https://aclanthology.org/N19-1423/
M. Phan and P. Ogunbona, “Modelling context and syntactical features for aspect-based sentiment analysis,” in Proc. Annu. Meeting of the Association for Computational Linguistics (ACL), Online, 2020, pp. 3211–3220. Available from: https://aclanthology.org/2020.acl-main.293/
W. Zhao, Z. Zhao, X. Lu, J. Wang, Y. Cheng, and A. Zeng, “Is ChatGPT a good sentiment reasoner? A preliminary study,” arXiv preprint arXiv:2304.09582, Apr. 2023. Available from: https://doi.org/10.48550/arXiv.2304.09582
R. Brooks, J. Kim, M. Patel, and S. Wang, “Evaluating Claude for affective computing: Constitutional AI and emotion sensitivity,” in Proc. Int. Conf. on Computational Linguistics (COLING), Torino, Italy, 2024, pp. 1821–1834.
X. Huang, Y. Zhang, L. Chen, and K. Wang, “Chain-of-thought prompting for emotion analysis with large language models,” in Proc. Conf. on Empirical Methods in Natural Language Processing (EMNLP), Singapore, 2023, pp. 1234–1248.
Y. Li, H. Yan, T. Liu, and X. Luo, “EmotionBench: A comprehensive benchmark for evaluating LLM emotion understanding,” arXiv preprint arXiv:2402.03042, Feb. 2024. Available from: https://ieeexplore.ieee.org/abstract/document/10595504
J. Zhang, W. Li, and Q. Chen, “Comparative analysis of cloud-based NLP deployment architectures for production workloads,” IEEE Transactions on Cloud Computing, vol. 10, no. 3, pp. 1892–1907, Jul. 2022.
R. Patel and S. Kim, “Best practices for deploying LLM inference services on AWS Bedrock,” in Proc. IEEE Int. Conf. on Cloud Computing Technology and Science (CloudCom), Naples, Italy, 2023, pp. 45–52.
Singh, V. Kumar, and T. Reddy, “Infrastructure-as-Code toolchain comparison for AI-powered API services on AWS,” in Proc. IEEE Int. Conf. on Cloud Computing (CLOUD), Chicago, IL, USA, 2023, pp. 123–130.
V. Kumar and R. Sharma, “Scaling emotion analysis APIs on AWS: From prototype to production,” ACM SIGOPS Operating Systems Review, vol. 58, no. 1, pp. 34–42, Jan. 2024.

Cites this article as

N. Singh "Detecting Emotions in Chat Messages with AI: A Real-Time REST API Built on AWS Bedrock", International Journal of Innovative Research in Engineering and Management (IJIREM), Vol-13, Issue-2, Page No-189-197, 2026. Available from: https://doi.org/10.55524/ijirem.2026.13.2.24