Automobile competitive product comparison method based on viewpoint mining analysis

Abstract

The invention discloses an automobile competitive product comparison method based on viewpoint mining analysis. The method comprises the steps of firstly constructing an automobile user evaluation training data set, and then training a text topic extraction model and an LSTM-Attention training text emotion polarity discrimination model by using a deep learning network structure including a convolutional neural network TextCNN; in combination with two deep learning models of TextCNN and LSTM-Attention, obtaining an automobile user evaluation theme and theme emotion scores by using a pipeline processing mode, and finally outputting theme evaluation and emotion polarity by using the model to realize automobile competitive product comparison based on automobile user comment viewpoints. Compared with the prior art, automobile competitive product analysis comes from user comments, the method has the advantages of being high in authenticity and feasibility, and the model has good accuracy anduniversality.

Description

technical field [0001] The invention relates to the field of viewpoint mining and analysis of natural language processing, in particular to a method for comparing competing automobile products based on viewpoint mining and analysis. Background technique [0002] With the rapid development of the Internet 2.0 era, great changes have taken place in traditional lifestyles and business structures. The Internet accompanies every aspect of everyone's life. People are accustomed to leaving a large amount of text information on the Internet to express their opinions and emotions. With the government's vigorous support for new energy vehicles and the rise of smart connected vehicles, both indicate the diversified development and transformation of the automotive industry in the next few years, automakers need to know whether their products can meet the needs of consumers, but traditional research methods are due to Defects such as small sample size and low efficiency have been unable ...

AI Technical Summary

Problems solved by technology

But the disadvantage is that there is a problem of long-term dependence. Using the self-attention mechanism allows the decoder to use the final hidden state and the information calculated based on the input hidden state sequence to effectively solve the problem of long-term dependence, and can capture each in the sequence. local key information

Images