Semantic Analysis Of Text Based On Emotional Classification

With the increasing popularity of the Internet, the Internet media is becoming an indispensable information media. Different from traditional media, Internet media, with the open dissemination of information, extensive, fast and so on. However, the information content of safety problems, which to the Internet for information security management with unprecedented difficulties. The Internet provides a wealth of information resources, has brought the proliferation of bad information. Internet pornography, violence, gambling, drugs and other illicit spread of harmful information will be spiritual civilization and the quality of people's health, especially for a bad impact on children's physical and mental health. Network of political attack, reactionary expression and information, on the country's stability and unity, the steady development of society has a very negative impact. Some information can not simply by keyword matching to determine whether the adverse information, such as the Universal Declaration of cult and critical information. Analysis of such information in the time, not only to analyze the information contained in the subject matter (topic), but also determine its position and attitude, that orientation (polarity).In this paper, harmful information on the commentary on the text of a study classified the emotional tendencies, mainly as follows:In this paper, information gain and latent semantic analysis method, based on information gain and latent semantic analysis of a mixture of text dimension reduction method based on all training data and test data, information gain and the Unified Modeling latent semantic analysis. Experimental results show that the method is feasible.In this paper, the defects latent semantic analysis, probabilistic latent semantic analysis using methods to construct the text-the words of co-occurrence matrix, using the EM algorithm to solve. Experimental results show that the same test set, based on probabilistic latent semantic analysis of text sentiment classification methods achieved good classification results.There are too many risks and network violations, serious impact on enterprise network security in internet, often causing irreparable damage. Importance of information security are rising, the information content of security management network security has become the new requirements. This design content security management system, experiments show, the system can avoid man-made network security risks, promote Internet usage, the network environment more secure, Conger use of network and create more value.