Development of an Automatic Classification Corpus for Business English Vocabulary Using Bidirectional Long Short-Term Memory Network

As the world economy progresses, automatic vocabulary classification is a crucial task to improve the effectiveness of information retrieval and Natural Language Processing (NLP) due to the widespread use of business English in various fields. There are many disadvantages to using traditional methods based on manual rules to process large amounts of dynamic text input, mainly ignoring the contextual information of vocabulary, resulting in low classification accuracy. This paper constructs an automatic classification corpus of business English vocabulary based on the Bidirectional Long Short-Term Memory (Bi-LSTM) network. It improves the classification precision and system robustness through deep learning technology to solve problems. Research methods include: First, a business English vocabulary dataset is constructed and preprocessed, covering data collection, cleaning, word segmentation, tagging and vocabulary screening. Second, a Bi-LSTM model is designed and implemented to capture the contextual information of vocabulary through bidirectional modeling, thereby improving the classification accuracy of polysemous words and domain-specific vocabulary. Subsequently, model training and optimization are performed, and cross-validation is used to adjust hyperparameters such as learning rate, number of LSTM cells and batch size to ensure optimal performance. Finally, the model is deployed, and the classification results are integrated through the Application Programming Interface (API). Experimental results show that the Bi-LSTM-based model achieves a storage accuracy of 99.3% and a processing efficiency of 549.5 words per second, significantly better than the traditional rule-based method (accuracy of 92.3% and processing efficiency of 384.6 words per second). In addition, Bi-LSTM shows substantial advantages in system scalability and robustness and can effectively cope with new vocabulary and context changes.