Multimodal discourse analysis of English reading instruction in colleges and universities based on weighted function algorithm

Abstract

Abstract This paper proposes a kernel clustering method that uses the weighted function method to address the problem of excessive similarity calculation in the kernel function clustering process. Based on function-based clustering with orthogonal basis expansion, function-based principal component analysis is used to reduce the dimensionality of function-based data and extract the first few principal components that can contain the most information of the original data. The factors of each component are assigned with characteristic weights so that the scores of the principal components replace the original functional data for clustering analysis of English reading teaching in colleges and universities. A systematic functional language framework was used to conduct multimodal discourse analysis based on clustering results. The symbolic annotation of pure language reached 14.5, and the symbolic resources of visual imagery reached 85.7 during the CB stage of English reading instruction. The number of semiotic resources was relatively small, so the complexity of the modality was lower than in other stages.