Analysis Technique of English Composition Themes Based on Artificial Intelligence Context Theme Recognition

Current English composition assessment mainly relies on manual scoring, which is time-consuming and highly subjective. For this reason, artificial intelligence technology is used in the study to propose a context based topic recognition model for analyzing English essay themes. In the contextual theme recognition stage, the model constructs a potential theme model using the Probase knowledge base and obtains the local semantic vectors of English compositions using the Bidirection Ecoder Representations from Transformers (BERT) model. The sentiment viewpoint analysis stage uses the Gated Recurrent Unit (GRU) neural network to obtain deep vector representations of the compositions. Finally, ratings and comments are generated based on topic relevance and sentiment viewpoint scores. The outcomes revealed that the model has high accuracy and F1 value on the tasks of topic recognition, topic extraction and text classification, which fully consider the contextual information of the text. By adjusting the relevance threshold, the model achieves 89.2% accuracy in topic analysis and 86.7% recall on 2000 English compositions. The outcomes revealed that the method is effective in English composition theme analysis. The study combines artificial intelligence neural networks to provide new ideas and methods for automatic evaluation of English compositions, which has practical value.