Analyzing Perceptions of Intelligent English Teaching Tools Using a Dirichlet–Multinomial Bayesian Machine Learning Model

With the rapid development of artificial intelligence and machine learning technologies, intelligent English teaching tools based on speech recognition and natural language processing have been increasingly applied in language education. However, systematic analysis of users’ perceptions and attitudes toward such technologies remains limited. This study aims to evaluate the perceived effectiveness of intelligent English teaching tools using questionnaire survey data. Since the dataset consists of aggregated categorical responses, a Dirichlet–Multinomial Bayesian model is employed to estimate the underlying probability distribution of Likert-scale responses and to perform probabilistic inference on respondents’ attitudes. Experimental results based on 315 valid responses show that most questionnaire items exhibit high posterior mean scores and strong probabilities of positive responses, indicating that respondents generally hold favorable views toward intelligent English teaching technologies. The findings suggest that AI-driven language learning tools are widely recognized as effective supplements to traditional English instruction. Furthermore, the study demonstrates that the Dirichlet–Multinomial probabilistic framework provides a practical machine learning approach for analyzing categorical questionnaire data, offering methodological insights for future research on AI-assisted education.