Detection And Analysis Of Public Security Events Based On Convolutional Neural Network

Earthquake,disease transmission,terrorist assault and other public security events will bring great loss to our society,and spread to the political,economic,scientific and technological fields.With the continuous improvement of Internet infrastructure,the continuous expansion of network scale and the gradual improvement of people's awareness of social participation,more and more people know the public security events through the Internet and participate in the discussion and dissemination of such events.When a public security event happend,extracting the important factual information in the public security events through analyzing the massive text data in the Internet,then constructing the corresponding topic network is significant to understand,control and guide the public security events.It is also helpful to maintain the harmony of the society,protect people's lives and property security in real life.Based on the background,this paper focuses on the three types of public security target events such as "attack","injury" and "death",and proposes a method to detect and analyze public security events based on convolutional neural network.Firstly,this paper introduces the related basic concepts and authoritative evaluation criteria of event extraction,and elaborates the general steps of event type recognition in event detection and recognition.Secondly,this paper proposes a method of detecting and analyzing public security events based on convolutional neural network.In order to achieve the public security target event set,this paper detects and identifies events from the Internet through using the key words and the text classifier.For the public security target event set,this paper obtained the named entity and their relationship by Chinese Academy of Sciences speech tag set.Using the open source software UCINET6 to build a visual topic network,this paper also makes the micro-analysis of the physical nodes in the topic network.Thirdly,the experiments based on real data evaluate proposed method in terms of correctness and validity.The results show that proposed method in this paper can effectively analyzing the public security event.Finally,the work of this paper is summarized,and the future research work is prospected.