InsightEdu: Mobile Discord Bot Management and Analytics for Educators

Modern educational environments increasingly rely on digital platforms to facilitate interaction between students and educators. Discord has emerged as a popular communication platform in academic settings, offering a combination of messaging and support for chatbot development. However, most existing Discord bots lack specialized educational functionalities and mobile-friendly interfaces, limiting their effectiveness for instructional use. This paper presents InsightEdu, an innovative iOS application that provides a touch-centric interface for managing a custom Discord bot designed for educational contexts. The system enables educators to conduct surveys, collect feedback, and track attendance through an intuitive mobile interface. The architecture combines a SwiftUI-based iOS client application with a Python-based Discord bot server. User evaluation with educators demonstrated significant usability improvements compared to traditional Discord interfaces, with 92% of participants (n = 20) reporting enhanced efficiency in managing educational interactions. This study demonstrates that mobile-first, instructor-friendly design can significantly enhance the utility of existing communication platforms for academic purposes.

💡 Research Summary

The paper introduces InsightEdu, a mobile‑first system that lets educators manage a custom Discord bot through an iOS application built with SwiftUI. The architecture consists of three layers: a SwiftUI client following the MVVM pattern, a Flask‑based REST API gateway, and a Python Discord bot implemented with py‑cord. By decoupling presentation from bot logic, the system preserves the pedagogical capabilities demonstrated in the earlier TUMDiscordBot prototype (attendance tracking, feedback collection, and multi‑step surveys) while eliminating the need for command‑line interaction and JSON configuration files.

Key functional requirements (FR‑1 to FR‑6) cover starting/stopping attendance sessions, creating simple and complex surveys, triggering feedback dialogs, persisting bot settings locally, and visualizing historical data with interactive charts. Non‑functional requirements (NFR‑1 to NFR‑5) enforce ≤ 300 ms round‑trip latency over campus Wi‑Fi, compliance with Apple’s Human Interface Guidelines, authenticated API access, support for multiple bot instances, and a roadmap toward JWT‑based security.

The iOS UI is touch‑centric: large hit targets, hierarchical navigation via NavigationStack, immediate visual feedback, and Apple Charts for data visualization. The Flask server runs a dual‑threaded model—one thread handles HTTP requests and authentication, the other maintains a persistent Discord gateway connection. Endpoints are organized as blueprints, follow Level 2 REST conventions, and return RFC 7807‑style error payloads. Data are stored locally as CSV files, enabling easy export to institutional analytics pipelines.

A user study with 20 instructors compared the mobile interface against traditional CLI‑based bot management. Participants performed six core tasks (attendance, simple/complex surveys, feedback, configuration, and historical analysis). Results showed a 40 % reduction in median task completion time, a drop in error rate from an average of 2.3 errors per task (CLI) to near zero (mobile), and a 92 % preference for the mobile app. Heuristic evaluation yielded an average score of 4.2/5, with aesthetic design scoring 4.8/5. NASA‑TLX scores indicated a significant decrease in perceived mental demand, confirming that recognition‑based touch interactions are less cognitively taxing than recall‑based command syntax.

Performance measurements on a Raspberry Pi 4 B (4 GB RAM) demonstrated that database queries averaged 180 ms, Discord API calls 250 ms, and overall 95th‑percentile round‑trip latency stayed below 300 ms, meeting the NFR‑1 target. Stress tests showed the system could support 12 concurrent instructors managing 20–50 students each, with memory usage stabilizing at 1.2 GB and CPU at 15‑20 %. Network traffic remained minimal (< 1 KB per request), and automatic reconnection recovered functionality within 30‑60 seconds after a network interruption.

The discussion highlights the benefits of building on a free, reliable platform (Discord) and low‑cost hardware, thereby reducing barriers compared to traditional LMS solutions. Architectural lessons include the trade‑off between simplicity (single‑process deployment) and scalability (future containerization), and the need to evolve from static API keys to JWT or SSO for broader institutional adoption. Limitations are acknowledged: a small sample size, lack of longitudinal data, and dependence on Discord’s API and terms of service.

Future work proposes three directions: (1) integrating large language models to generate actionable teaching recommendations from student feedback, (2) extending support to Android via cross‑platform frameworks such as Flutter or React Native, and (3) conducting semester‑long studies to measure actual student learning gains and engagement.

In conclusion, InsightEdu fills a critical usability gap by providing a touch‑centric mobile interface for managing Discord‑based educational bots. The system achieves a 40 % efficiency gain, near‑zero error rates, and strong instructor preference, all while operating on inexpensive hardware and leveraging Discord’s free infrastructure. The authors argue that mobile‑first design and operational simplicity are essential for widespread adoption of educational technology in diverse academic settings.