Study On The Model Of Public Crisis Warning In Microblogging Environment

To avoid crisis or eliminate it in the bud is the most economic and effective method of crisis management. The thought of Public crisis management still stay at the present: put too much focus on how to deal with the crisis, but ignore the prevention. That makes the crisis which can be originally found in the incubation period and eliminate by a series of effective public relations means to get rid of the crisis of development and finally became a serious social problem. The birth of the social network, led by weibo, allowing all citizens to express their views and opinions in a "zero threshold" way, and so, more and more social crisis event began brewing and outbreak in weibo. The current situation of the domestic study of public crisis warning in weibo environment is rare, most of them are focus on the results of crisis event, and the trend research after the crisis happened. The lack of prior preventive early warning research is the biggest problem, studies on preventive early warning indicators and the design of the model building are also stay in the traditional way, the lack of research under the network environment which is suitable for WEB2.0, and therefore it is necessary and feasible to explore and enrich the public crisis warning research based on the basis of predecessors’.The author build the model of public crisis early warning under the microblogging environment includes two sub models. Respectively is: the event pattern recognition model, and the index of diagnosis model. Event recognition model based on working principle is to use frequency domain analysis of weibo emergency monitoring technology to get the original data from sina weibo generated hot topics and choose those sudden public events which is likely to develop to the public crisis, prepare to choose the predicting object for the next work. Index diagnosis module is divided into two parts, first of all, from two levels(events, user), four aspects(weibo event strength, heat, weibo event events state of public opinion, public opinion development direction) I build eleven specific indicators(authenticity, sensitivity, weibo number, microblogging forwarding, thumb up, publishers the effect, view orientation degree, body, tolerance, turn public opinion change frequency, public opinion, public opinion change activity) public crisis early warning under weibo environment index system, and then by the BP neural network algorithm analysis and calculation in the indicators of the incubation period, the event to sufficient training samples, output for the outcome of the next step. Results output module is on the premise of fully trained, input relevant predictor of samples, to predict the development trend of next time point and the final output a warning state.The author on the basis of building the early-warning model, through the "dalian PX project mass incidents" verified the effectiveness of the model, this article main contribution is as follows:(1) the weibo environment emergency identification process into the process of early warning, and has carried on the empirical research on it.(2) proposed a new suitable for weibo environment public crisis early warning index system.(3) based on the analysis comparison algorithm, the BP neural network algorithm was applied to public crisis warning, weibo environment on the application of innovation.