A Review Quality Classification Model Based on TS-BiLSTM Oriented E-Commerce Platform
Comment quality classification can be used to select high-quality reviews,which are widely used in e-commerce and other fields.High-quality reviews can be an effective way for businesses and consumers to make in-formed decisions about their product choices.However,due to the characteristics of interlacing and dispersing of user reviews on e-commerce platforms,the process of feature extraction is complex,and the traditional quality classification of reviews generally adopts machine learning methods,so the accuracy of classification is not high.To solve this prob-lem,this paper proposes a user review quality classification model based on TS-BiLSTM(TinyBERT Self-Attention BiLSTM)oriented e-commerce platform.Firstly,TinyBERT is used to preprocess the comment text to construct the word vector,then BILSTM is used to extract the features of the input word vector,and Self-Attention is used to calcu-late the weight of the extracted feature vector.Finally,the weighted feature vectors are classified by full connection and Softmax.The experimental results show that the proposed model can effectively improve the accuracy of text qual-ity classification for e-commerce platform reviews.
NETL
NSTL
万方数据
