Design and Development of a Personalized Chinese Vocabulary Learning System using Fine-Tuned DeepSeek Language Models

With the development of globalization, learning Chinese vocabulary by non-native speakers becomes more and more crucial. However, traditional learning processes cannot keep pace with the needs of contemporary fast-paced learning because of inefficiency. Deep learning technology, a new data processing technique, offers novel means for acquiring personalized and efficient language learning. By utilizing this technology, the current study seeks to enhance the process of English word learning, most notably strengthening the mastery of technical vocabulary, bringing the revolutionary approaches to language teaching and testing their applicability. To fulfill these requirements, this research paper suggests the enhanced Large Language Model (FT-LLM) based deep learning structure for delivering the individual learning solutions depending on learner's given situations. This framework examines the learning history and behavior of the learners and can dynamically modify teaching content and difficulty, achieving precise vocabulary teaching alignment. The proposed framework is experimented with Kaggle's English-Chinese learning data and performance metrics like BLEU, ChRF, precision, recall are computed. The performance of the proposed architecture is compared with the other available learning frameworks. Evaluation outcomes prove the FT-LLM model has attained BLEU score of 0.99, ChRF Score of 0.22, precision of 0.988 and recall of 0.977.