News text classification based on attention mechanism

News text is often divided into two parts: title and content. In the research of news text classification, many studies have not considered the characteristics of news text classification. It is unscientific to merge the title and content as input hastily. The title is of great value to news text, and its importance is ignored in text classification, which has a certain impact on the classification effect. Aiming at this problem, this article designs a word vector weighting method based on attention mechanism, and builds a bidirectional long short memory neural network (BiLSTM) model for text classification. The experimental results show that BiLSTM news text classification with the introduction of attention mechanism improves the classification effect.