Creation of BAPOLAIC: An AI-Powered Chatbot for Document Recognition and Voice Assistance

Abstract

The swift progress of Industry 5.0 has considerably increased the need for intelligent, multimodal interaction systems in educational environments. Here, we discuss the design, realization, and appraisal of BAPOLAIC (Batam Polytechnic AI Chatbot), a web-based multimodal assistant that combines Optical Character Recognition (OCR), Natural Language Processing (NLP), and voice interaction technologies, supported by the Gemini API. The system adopts a three-tier client-server model to overcome the limitations of current academic chatbots by offering smooth document recognition and multimodal interaction. An extensive evaluation was performed to measure the performance of the components and the acceptance of the users. The performance metrics showed a very high reliability, including 98.7% OCR accuracy, 85.7% NLP retrieval success, 2.1% Word Error Rate (WER) for voice commands, and a 4.3/5.0 Mean Opinion Score (MOS) for the quality of the voice output. A System Usability Scale (SUS) evaluation gave a score of 58.0, which implies that the system is operational but also marks the need for user experience improvements in future. In this research, we provide a validated framework for the integration of multiple AI technologies into a specific educational tool, thus laying the groundwork for future intelligent academic assistance systems.

Ähnliche Arbeiten