A Smart Offline Educational Assistant for High School Students Using Flutter and SQLite: Design, Implementation, and Comparative Evaluation

Abstract

In regions with limited internet infrastructure, most intelligent learning technologies fail to deliver consistent access and localized content. This study presents Scientific Assistant, a fully offline educational application for Libyan high-school students, developed using Flutter 3.x (Dart), SQLite 3 (sqflite), and a TF-IDF engine with Arabic preprocessing. Following SDLC, the study covers requirement analysis, three-layer architecture, implementation, and evaluation. Four scientific subjects align with the Libyan secondary curriculum; the system supports dual student/teacher roles. Results: 0.9 s response time, Precision=0.93, Recall=0.91, F1=0.92, usability 9/10 (n=30, two-week UAT). The study demonstrates that curriculum-specific, Arabic-adaptive, fully offline AI systems can substantially enhance learning accessibility in resource-constrained environments.