Construction of Online Interactive Teaching Platform for College English Based on Multilayer Perceptual Machine Modeling

Abstract

Abstract Most of the existing English online teaching platforms are just digital replicas of offline teaching, which cannot fully utilize the advantages of information technology. Based on the learning theory of constructivism, this paper builds the functional architecture of a personalized English interactive teaching platform and completes the design of the main functional modules. According to the quantitative results of the platform’s user proposition requirements, a matrix description method for English test papers is proposed, and an algorithm is designed to realize the automatic generation of test papers. Meanwhile, the resource recommendation mechanism of the English platform is optimized using a multilayer perceptual machine. Unstructured English learning resources are processed based on the word vector representation method, and for the problem of insufficient processing of nonlinear data in the DN-CBR model, multilayer training is carried out using the MLP to improve the perceptual degree of this part of the data, and the design goals are verified through the test of the teaching platform. The correlation coefficient between self-assessment difficulty and topic difficulty is 0.368, and the correlation coefficient between self-assessment difficulty and positive error is 0.163, indicating that there is a moderate correlation between learners’ self-assessment difficulty and the difficulty of the system topics and between learners’ self-assessment difficulty and positive error. The MLP-based pushed questions’ difficulty is adapted to the learner’s personalization level by dynamically revising it based on their learning process data.