Car Review Mining And Application Based On Deep Learning

With the increasing popularity of technology and economy,people can easily access the Internet anytime and anywhere,it can produce a huge amount of Internet data.A large part of this data contains some subjective information such as people’s thoughts,concepts,emotions and evaluations.The mining of such information plays an important role in social and economic development.However,due to the unstructured and the volume of these data,it is very difficult to analyze these data.Based on the deep learning,the paper uses natural language processing and the internet data to automatically learn data statistics and implement it into engineering applications.At the beginning,this paper first expounds the origin and significance of the research content of this topic,and then introduces the content of the topic.This topic is a combination of algorithm research and engineering implementation,so the architecture and establishment process of the project are explained first.And a description of the relevant requirements analysis and application of the technical explanation.Then it describes the principle of the crawler technology combined with deep learning CNN,data preprocessing Word2 Vec technology,and a brief description of the network structure used for deep learning.Then the main algorithm work of this topic is introduced,including the proposed Chinese verification code verification model,and the data preprocessing method is linked in two directions.For the mining of text data,this paper proposes a classification network based on LSTM improvement,and proposes an LSTM model that adds a processing layer that weights the average of hidden states of each layer.Finally,experiments were carried out.The experimental results were verified in the data sets that were effectively manually labeled,self-generated datasets and online datasets.The comparison with traditional popular methods was also carried out.The methods proposed in this paper are verified and evaluated from many aspects.