Research On The Text Classification Method Based On The Customer Complaint System Of A Company

With the rapid development and wide application of the Internet,people have gradually entered the era of information overload from the era of information shortage.In the face of such an explosion of information,people usually cannot quickly locate valuable content.As the basic research in the direction of natural language processing,text classification has very profound research significance and wide application prospects.In recent years,deep learning has gradually replaced traditional machine learning method,and has gradually become the main research direction in the field of text classification.Deep learning integrates the process of feature extraction and classification,has replaced the traditional text classification method,and can automatically obtain the characteristic information of the data.Deep learning has a more prominent performance,especially in the fields of data mining,computer vision,natural language processing and so on.This paper mainly takes the customer complaint problem analysis platform system of Company A as the research object.The customer complaint problem analysis platform is a service platform for centralized management and handling of customer complaint problems,which mainly provides services for various products,platforms and components within the company,and provides test personnel with lists of problem for improvement.The list of problems is divided into two different categories: List of problems that can be handled and list of problems that cannot be handled.To solve the problem of text automatic classification in the list of problems,we carried out the research with convolutional neural network algorithm in deep learning.First of all,because of the particularity of Chinese text in the list of problems,we used the current mainstream J ieba word segmentation tool based on the P ython programming language in the word segmentation part of data preprocessing.And for the particularity of the research object field,we re-added and reorganized a series of words to the stop list,which has laid a good foundation for the follow-up work of the text classification task of the list of problems.Secondly,for the segmented text,we used Word2 Vec technology to convert it into the expression form of word embedding,and then it was introduced into the convolution neural network model in the form of two-dimensional matrix.Finally,in order to prove the effectiveness of the convolutional neural network model for text classification of the list of problems,in this paper,the Naive Bayes algorithm with better performance in text classification in traditional machine learning was introduced as a comparison.And the experimental results showed that the convolutional neural network model classifier performs well,which can overcome some defects of shallow machine learning in text classification,and improve the performance of text classification of the list of problems.In addition,according to the increasing data volume of customer complaint platform,the effect of convolutional neural network model classifier based on deep learning will continue to rise,and can reduce certain manpower investment after being put into use by Company A,with relatively important realistic value significance.