Evaluation and Analysis of English Predicate Constructions Incorporating Multiple Linear Regression Algorithms

Abstract

In this research, the multiple linear regression methods, and the grammatical characteristics of verbs are studied on regression matrices and logical relations, and a model for teaching English predicate constructions is designed by multiple linear regression matrices and arithmetic rules. The qualities of subject verbs as sentence focus points are discussed in this study, along with two methods for recognizing English predicate verbs that are based on attention mechanisms and BERT. The focus mechanism-based method has the potential to improve the dependent verb's recognition performance over the standard approach by extracting the phrase's long-distance semantic dependence information. The BERT-based English predicative verb recognition model improves upon the prior technique by making more effective use of the input corpus. A multiple linear regression-based strategy is suggested for distinguishing predicate verbs that are unique within phrases. By setting the classification fitting requirements during training, the information can be optimised during the punishment phase in terms of production and fully suit the collective identity of the conditional verb.