ARTIFICIAL INTELLIGENCE INTEGRATED E-LEARNING SYSTEМ

Abstract

The foundation of this study relies on the fact that English is the global language that connects people around the world. It also plays a significant role as a medium for communication in global interactions in various sectors such as academic settings, business environment, or international travel. At present, many companies and institutions require English proficiency as a basic skill for hiring. Being able to communicate well in English has become a crucial skill not only for career development but also for accessing information and international networks. However, many individuals in non-English speaking countries still face challenges in learning English as a language. These struggles often arise due to lack of practice and low confidence in real-life conversation. To address this issue, the objective of this research is to combine and utilize the advancement of technology, particularly in the field of Artificial Intelligence (AI) to detect and correct errors from learners. This project aims to improve speaking, writing, and understanding in English by instant feedback on grammar and sentence structure. The final system achieved an accuracy of 85% in identifying and correcting grammatical mistakes and inappropriate expressions. This result demonstrates the potential of Al-assisted tools to improve English language learning in an interactive, accessible, and personalized way. The methodology involves collecting comprehensive dataset, applying preprocessing steps such as tokenization, cleaning, and labeling, build a neural network architecture such as Computer Neural Network (CNN) and Bidirectional Encoder Representations from Transformer (BERT), and Robustly Optimized BERT Pretraining Approach (ROBERTa) to detect and correct grammar error and hate speech detection. The system was able to achieve 85% accuracy in finding and fixing grammar errors and incorrect word choices. The overall program achieved 90% of accuracy in detecting both grammar mistakes and bad words. These results highlight the potential of technological solutions to support language learning.