Design and Development of Cross-Cultural English Communication Teaching Systems Based Hybrid Learning Algorithm

Language is considered to be a powerful tool that connects the versatile cultures across the globe, thereby preserving their own customs and traditions. It is the best communication tool for imparting education even to the last man of society. Hence, an intelligent framework is required to extract the essence of the multiple languages and translate into English language, which would be essential to prepare the heritage-rich contents. In recent times, Neural Machine Translation (NMT) is used for translating the language texts from source language to target language, which enhances the user's ability to understand the other linguistic features. However, translating the low-resource remains to be complex bottleneck which prevents the existing education systems to penetrate the heritages of the other languages into their teaching academics. In this context, this research paper proposes the hybrid combination of Long Short-Term Memory and transformer-M2M model to translate the resource-constrained languages to English that aids for the better teaching process. The proposed framework is trained with Multi30k multimodal translation datasets. Consequently, performance of proposed model is calculated using Bilingual Evaluation Understudy (BLEU), character-level F-score (chF) and Word Error Rate (WER). To prove excellence of proposed model, extensive comparison between the proposed and existing architectures is compared and its performance metrics are analysed. Experimental results demonstrates that proposed architecture has achieved average of BLEU score of 0.94, chF of 0.9, and WER of 0.10 in converting the other languages to the English. The proposed network has shown the new paradigm in teaching systems for the students to explore the cultures and traditions of the other languages that improves the mental ability of the students.