One Model Of Sensitivity Prediction Of Multilingual Text Message

Internet public opinion affects the stability and development of the region largely.It is an important issue that the system could discovery, analyze online publicsensitive opinion and responding quickly at the moment. The traditional method ofmonitoring of sensitive messages emphasis on the semantic understanding of textinformation, and text information on additional factors are not in consideration, andtechnology is more complex. Experiment take the time and affective into account, theadditional factors such as the latest in the forum, with bigger hits, response rates,speak registered users closer to the possibility of having a great sensitivity, so theyneed to be monitored in a timely manner to prevent it brings greater harm.Experiments were based on decision tree and BP network modeling approach to builda public opinion monitoring model of the monitor of sensitivity information, andprovide translation capabilities for the collected articles and these Uyghur texts couldbe translated into Chinese text, in order to more convenient, saving time and effort tomonitor public opinion on the network information.There are various methods for sensitive detect, the common method is a textstring pattern matching method with alignment algorithm based on a sensitivevocabulary, which determines whether the text message contains sensitive words.However, the correct rate for sensitive posts using this method on determining still islower, and it is to be improved. In this paper, two ways of Uyghur language of thesensitivity of the information to discriminate, one approach is to make datacollection, pretreatment, dig out the dimensions of the property and the sensitivity ofthe language relations by using the C4.5decision tree classification algorithm; anotherway is for data acquisition, preprocessing, using BP network to establish thesensitivity of the discriminate model, and gives the experiment and experimental dataanalysis and comparison of the two models their efficiency and reliability. Thetranslation platform of Uyghur and Chinese is the Microsoft Translator Hub, Parallel corpus learning, training, and ultimately can translate into Chinese, and get acorresponding Chinese text through the platform.Finally, the paper summarizes the processing technology in experiment and thetechnology of the field in the future and proposes appropriate countermeasures.