A Study Of Early Warning System Of New Media Events Based On New Genetic Algorithm

Currently, China is in the stage of social transformation and economic transition.Various social contradictions increased, under the historical conditions of accelerating thereform and opening up and establishing a new system of socialist market economy. In recentyears, with the increasing number of China’s Internet and mobile phone users, various socialcontradictions have transferred from the local discussion in the real world to the globaldiscussion in the network virtual world. We called these events that appear on the networkas the “new media event”. From SARS to7.23EMU accident, it has become a ubiquitousnetwork and social phenomenon.Early warning system of network is mainly aimed at the network information security,such as the monitoring of network intrusion and attack behavior, and less attention to theearly warning of the new media events. So the research on early warning system of the newmedia events is needed. Research the propagation characteristics of new media of “grouppolarization”,“Spiral of Silence”,“Butterfly Effect” and “boiling water effect”; found thatnew media events had two sides. Therefore, when a New media event occurs, relevantdepartments can’t simply laissez-faire development, nor can do a full denial. However, dueto the network information is many and diverse, it’s impossible to know each informationand make reasonable judgment to each information that be understood. Therefore, earlywarning system will need to have information automatically collected and classificationfunction. At this time, need according to former new media event to establish a clusteringmodel, and the clustering algorithm is the core part of the clustering model. For the twomain functions of early warning system, this paper mainly focuses on the followingrespects.1. New media event information collection. on the basis of the web crawler technology,build the website backstage automatic upload function, if the information meet theconditions, it is uploaded to the early warning system timely. This can fill the omission ofmanual collection or remove link of manual collection..2. The improvement of genetic algorithm. Based on analyzing the clustering algorithmusually used, the genetic algorithm and K-means are combined effectively. A clusteringfitting new media event that called Genetic Algorithm with K-means is proposed. The slowconvergence of genetic algorithm is improved, and the problem of easy to fall into local optimal solution of K-means is improved. The operations of optimal preservation strategies,single-point crossover and single-point mutation are used; it can ensure the convergence ofGenetic Algorithm with K-means.3. The realization of the function of warning. On the basis of the new media eventswhich have already happened to create a cluster model, then determine the type of the newmedia events. Design a set of basic processing measures for each type of new media event.The system will classify the collected new media event by classifier of Bayesian network,and make a right processing, in order to achieve early warning to the new media event andprevent the occurrence of group incidents. Simulation analysis demonstrated the feasibilityand the effectiveness of the algorithm.