Smart Vocabulary Builder: An AI-Driven Personalized Learning System Using Spaced Repetition and Generative AI

Abstract

The acquisition and long-term retention of new vocabulary are frequently impeded by traditional, passive learning methods like rote memorization, which fail to address the cognitive principles of memory and contextual understanding. This project, the Smart Vocabulary Builder, presents a full-stack, AI-powered web application designed to overcome these challenges by creating a highly efficient, personalized, and interactive learning environment. The system is architected with a robust Python-based FastAPI backend, a dynamic React.js frontend, and leverages a database managed by SQLAlchemy. At its core, the application implements a Spaced Repetition System (SRS), utilizing a priority queue to intelligently schedule word reviews at optimal intervals, thus combating the natural forgetting curve. Innovation within the project is driven by the deep integration of the Google Gemini API, which transforms the learning process into a conversational experience. When a user adds a new word, the application initiates an \"AI Consultation,\" providing not just a definition, but also a contextual example sentence and a creative mnemonic. This interaction is dynamic, allowing users to regenerate explanations if clarity is not achieved. The platform further personalizes the learning journey by proactively suggesting new, relevant words based on the user\'s existing vocabulary. A key feature is the integrated, browser-native \"Listen & Practice\" tool, which combines text-to-speech and speech-to-text functionalities for pronunciation feedback without reliance on costly external services. The result is a feature-complete, intelligent learning platform that successfully merges fundamental computer science algorithms, modern full-stack development practices, and the power of generative AI to offer a demonstrably more effective and engaging alternative to conventional vocabulary-building methods.

Introduction

Vocabulary acquisition is essential for language proficiency, influencing reading, writing, communication, and overall learning effectiveness. Traditional methods such as rote memorization, flashcards, and dictionary-based learning often result in poor long-term retention and limited contextual understanding because they do not incorporate adaptive learning or memory retention principles.

To overcome these limitations, the proposed Smart Vocabulary Builder integrates Spaced Repetition Systems (SRS) and Generative AI into a unified, intelligent learning platform. The system is designed as a full-stack web application using FastAPI, React.js, and a SQL-based database, providing a scalable and responsive learning environment. It offers AI-generated definitions, contextual examples, mnemonic aids, personalized vocabulary recommendations, and pronunciation practice through Text-to-Speech (TTS) and Speech-to-Text (STT) technologies.

Existing vocabulary learning platforms such as Anki, Quizlet, Duolingo, and Memrise provide partial solutions but suffer from several limitations, including lack of personalization, limited contextual explanations, passive learning approaches, and fragmented learning experiences. These systems generally fail to combine adaptive review scheduling, contextual understanding, pronunciation training, and intelligent content generation within a single platform.

The primary objectives of the Smart Vocabulary Builder are to:

• Improve long-term vocabulary retention through adaptive SRS scheduling.
• Enhance contextual understanding using AI-generated explanations and examples.
• Provide personalized learning paths based on user performance.
• Improve pronunciation through speech-based practice and feedback.
• Recommend relevant vocabulary according to learning progress.

The system includes modules for vocabulary management, AI-powered word explanations, adaptive review scheduling, personalized word recommendations, and pronunciation practice. Vocabulary data is processed and stored efficiently using SQLAlchemy, while JWT authentication ensures secure user access. Integration between the React frontend, FastAPI backend, Generative AI services, and speech technologies creates a seamless and interactive learning experience.

Testing is performed at multiple levels, including unit, integration, functional, system, performance, security, and user acceptance testing. Results indicate that the platform successfully provides reliable performance, secure data management, personalized learning, and improved vocabulary retention.

Conclusion

The Smart Vocabulary Builder presents an effective solution to the limitations of traditional vocabulary learning methods by integrating Artificial Intelligence with adaptive learning techniques. The system successfully combines a Spaced Repetition System (SRS) with Generative AI to enhance vocabulary retention, contextual understanding, and pronunciation skills. By providing personalized learning experiences, dynamic content generation, and interactive features, the platform transforms vocabulary acquisition into an engaging and efficient process. The implementation using a full-stack architecture ensures scalability, performance, and seamless user interaction. Overall, the proposed system demonstrates the potential of AI-driven educational tools in improving learning outcomes and offers a strong foundation for future advancements such as gamification, advanced analytics, and mobile application integration.

References

[1] FastAPI Documentation, “FastAPI Framework for Building APIs,” Available: HTTPS://FASTAPI.TIANGOLO.COM/ [2] React Documentation, “React: A JavaScript Library for Building User Interfaces,” Available: https://react.dev/ [3] SQLAlchemy Documentation, “SQLAlchemy ORM and Database Toolkit,” Available: https://www.sqlalchemy.org/ [4] Google AI, “Gemini API Documentation,” Available: https://ai.google.dev/docs [5] P. Wozniak, “Optimization of Learning Using Spaced Repetition,” SuperMemo, 1990. Available: https://www.supermemo.com [6] Mozilla Developer Network, “Web Speech API (Speech Recognition and Synthesis),” Available: https://developer.mozilla.org/en-US/docs/Web/API/Web_Speech_API [7] J. D. Kelleher, B. Mac Namee, and A. D’Arcy, Fundamentals of Machine Learning for Predictive Data Analytics. MIT Press, 2015. [8] S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997. [9] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT Press, 2016. [10] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. NAACL-HLT, 2019, pp. 4171–4186. [11] T. Mikolov et al., “Efficient estimation of word representations in vector space,” in Proc. ICLR, 2013. [12] A. Vaswani et al., “Attention is all you need,” in Advances in Neural Information Processing Systems, 2017. [13] D. Jurafsky and J. H. Martin, Speech and Language Processing, 3rd ed. Draft, 2023. [14] F. Chollet, Deep Learning with Python. Manning Publications, 2017. [15] C. D. Manning, P. Raghavan, and H. Schütze, Introduction to Information Retrieval. Cambridge University Press, 2008.

Copyright

Copyright © 2026 Dr. C. Udhaya Shankar, Elamurugan R, Lochana T. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.