The Research And Application Of The New Event Detection Method In Social Network Based On Hybrid Models

The traditional detection of new events tend to focus only on a specific link in the process of detection,but not construct a complete and comprehensive system that the new event detection accuracy and efficiency.Based on the existing study of new event detection,aiming at the existing social network testing new event monitoring nodes leads to more algorithm the efficiency is low,water users have an impact on the results,the results of a single topic,the existence of pseudo public opinion time accuracy is not enough even in the final result,in addition to the proposed how to efficiently obtain the core node,also consider the identification and Navy account based on the information the theme of community division,pseudo public opinion verification testing,aims to establish a relatively complete and practical,rigorous and comprehensive new Incident detection system.This paper takes Sina micro-blog as the research object.Firstly,EBKND algorithm is used to capture the core of user data,obtaining an initial database.Then,in order to ensure the accuracy and rigor of the new event,this paper introduces the culling of water army and theme community division.The culling of water army prevents the false users;In addition,Sina micro-blog has a more serious problem,it is about the entertainment news and events related to stars occupy a high attention,through the theme of community division,we can ensure the diversity and efficiency of the new event.The last step,the detection of pseudo events can eliminate the "false new event" in result,ensuring the accuracy of the final results.In the last chapter,this paper uses NED test and evaluation,from four aspects to evaluate our algorithm:loss rate,false alarm rate,the average accuracy rate and average detection time.Through calculation and comparison of these data,we can find the authoritative users influence analysis model for NED effect,EBKND algorithm the Navy and identification of two algorithms to ensure the accuracy and efficiency of the new event detection.Finally introduced the theme of community in NED are good,in addition to reducing the time complexity of the algorithm,the most important is to improve the diversity of topics.