| As the internet information rapidly growing, the harmful information, such as obscene information, violence information, reactionary information, and so on, has also swift growth. In all the means of spreading internet information, video is becoming the main one because of its characteristics such as intuitiveness and attraction.To face the new circumstances, discovery and screening the large amount of negative information by human becomes unrealistic. So we should get help from computer. The characteristics of video information is high storage cost, difficult to distinguish, and inefficiency for distinguish. And all most of video's text and label can reflect the video's content truly. Accordingly, the main method to analyze the harmful information is analyzing the text and label of the video.The main technologies of this paper are the support vector machine and N-gram model. With the analysis of the internet video information, internet information inspectors can get accurate and effective help, and also want to explore the natural language processing applications on the Internet. With the crawler, the system can get the information from web. With Chinese word segmentation, the system can pretreat data from web. With text categorization and clustering, the system will analyze the video information.The implementation of this paper depends on SNLP(Statistics Nature Language Processing). The results of the crawl will be word processing, support vector machines for text classification, to determine the extent of harmful Internet information. Through video monitoring of view the extent of its harm to society, the system will provide a right basis for decision making.The system to promote the strengthening of supervision of Internet information and organize forces to start sorting and in-depth analysis of information, emergency response network of public events, a full picture of social conditions and public opinion have a positive role in promoting. |
PDF Full Text Request