An Innovative Strategy Towards Oral English Assessment Using Machine Learning, Data Mining, and Blockchain Techniques

Abstract

Abstract Few studies have looked at how EFL students perform while speaking in public in English, despite the fact that video data analysis is often used in applied linguistics. In a world that is becoming more globalised, effective public speaking is essential, but EFL learners struggle to execute it, and despite the significance of characteristics like eye contact and speech pauses, there is currently no objective analysis of these elements. In-depth research and analysis on the topic of adaptive proofreading of English pronunciation in a wireless sensor network setting are presented in this article. Data mining refers to the method of generating knowledge and information from a significant amount of imperfect, unpredictable, fuzzily, and randomly generated data that was unknown prior but may be beneficial. This study examines how well EFL learners do orally using video data samples and a Machine Learning (ML) strategy known as a "discrete three-layered fuzzy logic artificial neural network (DTL-FL-ANN)". The architecture of the integrity system for college students is very compatible with the decentralization, security, trustworthiness, and traceability of Blockchain Technology, which is being used in many facets of life. The spoken English data was gathered and filtered for this research using the normalization method, median filter, and "Local Contrast Stretching (LCS)" procedures. The feature extraction stage employs the multi-manifold discriminant analysis (MMDA) to extract the significant attributes from the preprocessed data. We evaluate spoken English performance using the proposed DTL-FL-ANN+BT method. In order to maximize effectiveness, DTL-FL-indicators ANN+BT's are researched and contrasted with existing approaches. The suggested methods improved performance metrics such as accuracy (95%), evaluation error (55%) and evaluation time (45 sec). The proposed approach is determined to be the most effective when compared to the effectiveness of the current techniques for oral English evaluation using machine learning and blockchain technology.