• Amany Alsamhan

  Author

• Fatemah Hamad Aldosari

  Author

Artificial intelligence (AI) has been a significant asset in improving multilingual communication, allowing for accurate and efficient translation across a variety of languages. However, maintaining semantic meaning and the intent for communication in the context of a translation can be very difficult, especially in the field of education and cross-language communication. This study introduces an AI-Enabled Semantic Meaning Preservation Framework (AI-SMPF) for assessing AI-assisted translation with the parallel dataset of FLORES-200. The proposed framework is comprised of text preprocessing, AI-assisted translation, semantic representation, semantic similarity computation, meaning preservation evaluation, and translation quality evaluation. The framework measures semantic consistency not only based on lexical matching but also on the Semantic Similarity Score (SSS), Context Preservation Score (CPS), and the Translation Quality Index (TQI). These semantic evaluation metrics can also be used to facilitate automated feedback and guidance in education by objectively evaluating students' translation skills and context understanding, in multilingual learning activities. The experimental results show that the proposed framework obtains an SSS of 0.95, CPS of 0.93 and an overall TQI of 0.94, which shows that semantic meaning is preserved in multilingual translations. The statistical analysis shows that the average translation quality is 94.0%, with a standard deviation of 1.0%, which indicates that the quality of the translations is consistent across the different sentence pairs. The results show that using a semantic-aware evaluation is a more reliable method for evaluating AI-assisted translation than traditional lexical comparison-based methods. The proposed framework provides an efficient approach to maintaining the contextual meaning and facilitating multi-lingual communication in the field of English-language learning, which in turn helps to achieve more accurate AI-assisted translation for academic, educational and cross-cultural communication.

• PDF