Scholarlink

This paper introduces an innovative automated timetable management system to overhaul school scheduling, addressing issues like human errors, delayed updates, and limited access. Integrating Flutter for cross-platform mobile apps, a fine-tuned T5 transformer model for data processing, and Firebase for real-time cloud synchronization, the system ensures scalability and ease of use. Its standout feature is the automatic conversion of Excel (XLSX) schedules into structured JSON using natural language processing. Trained on a synthetic dataset of 10,000 diverse schedule samples, the T5 model achieved a 98.50% mean validation accuracy over 50 epochs, with a BLEU score of 67.07% and a 77.79% exact match rate, surpassing traditional rule-based methods. The system processes files in 5 s on standard hardware, optimized to 2 s with GPU acceleration, and scales linearly with timetable complexity, suiting institutions of all sizes. Future enhancements include multilingual support, reinforcement learning for optimized scheduling, secure offline functionality, and advanced analytics for resource planning. This research advances ed-tech by demonstrating how mobile development, AI, and cloud computing can resolve administrative challenges, offering a robust, user-friendly solution for modern school timetable management with significant potential for further innovation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI-Driven Timetable Automation Using Flutter, Firebase, and the T5 Transformer Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Lakkakula Saiakash,
  • Amar Jukuntla,
  • Kotha Pavankalyan,
  • Paladugu Mohanasai,
  • Nishith Kotha

摘要

This paper introduces an innovative automated timetable management system to overhaul school scheduling, addressing issues like human errors, delayed updates, and limited access. Integrating Flutter for cross-platform mobile apps, a fine-tuned T5 transformer model for data processing, and Firebase for real-time cloud synchronization, the system ensures scalability and ease of use. Its standout feature is the automatic conversion of Excel (XLSX) schedules into structured JSON using natural language processing. Trained on a synthetic dataset of 10,000 diverse schedule samples, the T5 model achieved a 98.50% mean validation accuracy over 50 epochs, with a BLEU score of 67.07% and a 77.79% exact match rate, surpassing traditional rule-based methods. The system processes files in 5 s on standard hardware, optimized to 2 s with GPU acceleration, and scales linearly with timetable complexity, suiting institutions of all sizes. Future enhancements include multilingual support, reinforcement learning for optimized scheduling, secure offline functionality, and advanced analytics for resource planning. This research advances ed-tech by demonstrating how mobile development, AI, and cloud computing can resolve administrative challenges, offering a robust, user-friendly solution for modern school timetable management with significant potential for further innovation.