MindTrack Journal: A Privacy-First Intelligent Journaling System with Emotion Detection

Abstract

MindTrack Journal is a secure, privacy-first, and intelligent journaling application designed to empower users to express their thoughts while maintaining complete control over their personal data. In a world where data privacy is increasingly important, MindTrack Journal stands out by ensuring that all user data remains private, encrypted, and never shared with third-party services. The application provides a seamless and user-friendly experience for managing personal journals, including core features such as user registration, login, and profile management, along with a rich block-based text editor supporting multiple content types for a more expressive writing experience. What differentiates MindTrack Journal from traditional journaling platforms is its intelligent analytics layer powered by a selfhosted AI service. Each journal entry is automatically analysed for sentiment using state-of-the-art transformerbased Natural Language Processing (NLP) models— specifically, finiteautomata/bertweet-base-sentimentanalysis for sentiment classification and bhadreshsavani/bert-base-uncased-emotion for emotion detection— ensuring that sensitive user data is never transmitted to external services. The results are visualised through an analytics dashboard, offering users meaningful insights into their emotional patterns over time. Security and privacy form the core pillars of the system, implemented via JWT-based stateless sessions validated by Keycloak, BCrypt password hashing, HTTPS/TLS encryption, database encryption at rest, and application-level journal content encryption. The architecture follows the Citadel Architecture pattern, combining Spring Boot 4.x Spring Modulith for core domain workflows with supporting microservices for analytics, payments, notifications, and ML processing. The system is implemented as a full-stack monorepo comprising a React 19/TypeScript frontend, a Java 21/Spring Boot 4.x backend, and a Python/FastAPI ML service, all containerised using Docker and orchestrated via Docker Compose, enabling repeatable deployment with a single command.