The Method Of TV Text Content Feature Extraction And Filtering For Triple Play

With the development of triple-play service, the channel of information and networkhas become more diverse and complex. Spreading of healthy and safe content is alsoincreasingly prominent. Currently, the mode of transmission makes illegal content escapeeffective regulation from regulators. In this case, content monitoring and the security ofnetwork information are increasingly important. In order to achieve the security ofinformation after the triple play, and promote the rapid development of text, video, images,audio and other services, following the design of principle of openness, operating,manageability, security, reliability, scalability and adaptability, achieving the administrable,controllable and operational goal, independently developed HUSTRIM which is thecontent monitoring system of TV based on triple play.Text monitoring subsystem of HUSTRIM can characterize the contents of the text,and filter the illegal textual content, so as to achieve the purpose of the textual contentmonitoring. In order to improve the rate of illegal text recognition and real-time, thesystem has developed the technology of the textual extraction and filtering. It hasimproved the traditional vector space model. The design and implementation of thestructure and modular function of the text monitoring subsystem has been detaileddescribed.Text monitoring includes text preprocessing, feature extraction of text and textfiltering. Text preprocessing is the basis of processing the text, and after receiving thestreaming media from feature classification system of HUSTRIM, achieving textpreprocessing by the word segmentation and removing the stop words. The featureextraction of text is responsible for feature extracting and analyzing of the text which hasbeen preprocessed. Text filtering matches the features of the textual tests with the illegalfeatures of the text to determine whether the text content is legal.Experiments show that the monitoring of the system is effective. The improvedmethod of text filtering can filter illegal text more effectively than the previous methodand also the efficiency of the text monitoring has been improved.