Font Size: a A A

Classification And Differences Of Chinese And American News Texts Based On Convolutional Neural Networks

Posted on:2020-06-05Degree:MasterType:Thesis
Country:ChinaCandidate:C N ZhouFull Text:PDF
GTID:2428330578955071Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the advent of the era of big data,the acquisition of information has become simple,but mining valuable information in massive data has become a new challenge.As the important channel for us to obtain information,the news media is of great significance in the era of big data.With the deepening of deep learning research,it has achieved outstanding results in the fields of image recognition and speech recognition.Therefore,applying deep learning to the field of text classification has important research and application value.This paper introduces the development history and research significance of text categorization,discusses the application of deep learning algorithm in the field of text categorization,and introduces the convolutional neural network algorithm in detail.Aiming at the problems faced by the current text analysis field,this paper proposes a news text classification algorithm based on convolutional neural network,and compares the differences between Chinese and American news media reports.The main research contents are as follows:1.Use crawler technology to continuously crawl news content published by major news portals in China and the United States.According to the definition of sociology,the news is divided into 32 categories according to the content,and the part of collected news is manually labeled as a training set.Compared to the news corpus obtained directly from the Internet,the news data sets used in this article are diverse and time-sensitive.It not only proves the performance of convolutional neural networks in multi-category text classification tasks,but also plays an important role in subsequent research.The network model trained in this paper has an accuracy rate of 83%in Chinese news texts and 90%in English news texts.2.Use the trained convolutional neural network model to predict the unlabeled news that is crawled.Then,the information entropy-based method,the topic model-based method and the sentiment analysis method are used to analyze the prediction results,and compare the differences between the news reports reported by the Chinese and American news media.The study found that the news reports in China and the United States have similarities and essential differences.In general,the Chinese media reported a large number of entertainment news,while the United States reported more crime-related news.In terms of the attitude of news reporting,the Chinese media is more active and the US media is more rational.For the same news category,the media of the two countries also have certain differences in the selection of news topics,which reflects to some extent the cultural differences between the two countries.
Keywords/Search Tags:Text classification, Convolutional neural network, Topic model, Sentiment analysis
PDF Full Text Request
Related items