Research On Evolutionary Analysis Methods Of Social Network Events

In recent years,with the development of Internet technology,people have become more and more connected to each other through social networks.The emergence of tools such as Weibo and Twitter has made it easier for people to understand the world.The social media platform generates a large number of messages every day,which contain a large amount of event information.Using topic detection and tracking technology to analyze social network events,tracking the evolution process of events is of great significance for grasping the development of events.Event evolution analysis is an aspect of topic detection and tracking technology,which mainly studies the event evolution process under the topic.This article will examine the evolutionary relationship between events in Twitter.Most of the traditional event evolution analysis methods first determine the time series relationship between events according to the time of occurrence of the event,and then calculate the event correlation degree to get the event evolution graph.In practice,the event may have noise,which leads to the inaccurate estimation of the event time,affecting the quality of the event evolution map,and thus affecting the subsequent event evolution tracking.In response to the above problems,thesis puts forward a new method of event evolution analysis based on the work of the predecessors.The main work is summarized as follows:(1)Thesis proposes an event temporal relationship analysis method based o n dynamic time warping algorithm.The method is aimed at the possible noise in the social network event.Firstly,the time series of the event is denoised by the kernel density estimation method.Then,the dynamic time warping algorithm is used for the time series of the denoised event and obtains the time alignment relationship of the two events;finally,it determines the timing relationship of the event.Experiments show that the proposed method can effectively obtain the timing relationship between events.(2)Thesis proposes an event evolution tracking method based on time division.This method is aimed at the evolution process tracking problem in the event evolution graph.Firstly,the event is divided according to the time window to form multiple stages of events.Then,the directed edges with weights are constructed for the events in the adjacent phase,generating the event evolution graph.Finally,according to the different target events,the Viterbi algorithm is used to solve the optimal path,and multiple evolution paths are organized into one or more evolution trees as an estimation of the event evolution process.Experiments show that the proposed method can effectively track the evolution process of event.