Page 168 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Journaling System accepts submissions from student users, following which the process
continues. The system stores the entry within its secure database to ensure information integrity
as well as preserve its confidentiality. The sentiment analysis model identifies core emotional
themes from each journal entry, with a target classification accuracy above 85%, emotion-specific
precision/recall metrics, and a false-positive rate below 5% for high-risk emotion detection,
ensuring timely and reliable feedback within a response time SLA of under 3 seconds. In
cases where high-risk emotional signals such as suicidal ideation or self-harm are detected, a
crisis detection module is activated. This module triggers a real-time escalation workflow that
includes alerting campus counsellors or emergency contacts based on predefined, consent-
based protocols and providing users with immediate access to in-app crisis resources such
as helplines and support messages. User emotional trends on the Mood Tracking Dashboard
get refreshed by the newly obtained sentiment data. The Affirmation & Insights Generator
is provided with the new patterns from the platform before initiating an inquiry for targeted
feedback. Once they provide information to the generator, it forms personalised affirmations
and suitable coping mechanisms in alignment with the emotional state of the users. The user
obtains blended synthesised feedback from the loop in the system, which includes emotional
trends merged with affirmations and mental health observations for achieving greater emotional
awareness and resilience.

5 References

[1] Vashisth, A., Kumari, M., & Mishra, A. (2024, September). MoodSync: An AI-Powered
Journal for Enhanced Emotional Well-Being. In 2024 International Conference on Artificial
Intelligence and Emerging Technology (Global AI Summit) (pp. 759-764). IEEE.
[2] Mishra, A. K., Bhartiy, K. K., Singh, J., Aluvala, S., Singh, P., & Kishor, K. (2023, September).
Unlocking the power of natural language processing through journaling with the
assistance. In 2023 3rd International Conference on Innovative Sustainable Computational
Technologies (CISCT) (pp. 1-5). IEEE.
[3] Angenius, M., & Ghajargar, M. (2023, July). Design principles for interactive and reflective
journaling with AI. In Science and Information Conference (pp. 62-81). Cham: Springer
Nature Switzerland.
[4] Angenius, M., & Ghajargar, M. (2022, November). Interactive Journaling with AI: Probing
into Words and Language as Interaction Design Materials. In International Workshop on
Chatbot Research and Design (pp. 150-170). Cham: Springer International Publishing.
[5] Valarmathi, R., Yazhini, G., Narmadha, L. S., & Janani, A. M. (2024, October). Enhancing
Mental Well-being through AI-Driven Virtual Environments: The Merrytopia Project. In
2024 International Conference on Power, Energy, Control and Transmission Systems
(ICPECTS) (pp. 1-6). IEEE.
.
.
[6] Vellore Institute of Technology, Chennai, India, https:// vit ac in/ campuslife/ studentswelfare
[7] RoBERTa, https:// huggingface .co/ docs/ transformers/ en/ model _doc/ roberta
[8] EmpathBERT: A BERT-based Framework for Demographic-aware Empathy Prediction,
https:// doi .org/ 10 .48550/ arXiv .2102 .00272

132