Abstract
The most important thing of Chinese short text classification is to extract and represent the features of short text. In the traditional representation methods, there are some problems, such as short text length and sparse features. Therefore, a short text classification model based on improved bilstm is proposed. The model includes word vector input layer, bilstm layer and attention layer. In the input layer, the pre trained word vector is used to transform the original text, and the intermediate vector representation with context timing information is obtained through the two direction (long-term and short-term) memory network; In the attention mechanism layer, the forward and reverse features are fused and weighted to obtain the short text vector with deep semantic features. The model is compared with other baseline models, and the experimental results show that its accuracy is significantly improved.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call