Research On Topic Mutation Monitoring Based On Dynamic Evolution Of Community Structure

As an important part of identification and evolution trend discoverying in disciplinary research,topic evolution analysis has attracted extensive attention of researchers.It can help researchers understand the deveolpments and trends in the field of discipline,to provide effective reference for strategic decision-makers.The existing researchs mostly make comparative analysis of the evolution between sequent time stamps using literature information,which can get the development and situation of discipline,but there exits a certain time lag.Further research on tracking topic evolution process is necessary to realize the theme mutation identification and monitoring,On the other hand,the analysis of the causes of evolution can be used to analyze the theme evolution law and the internal mechanism of the theme mutation formation.Most of the existing researches are focused on the topic evolution process and topic mutation identification,the mechanism of topic evolution and mutation formation still needs further perfection.To solve these problems,this paper proposes a method of topic evolution monitoring based on dynamic evolution of community structure.Combined with the critical point of community structure change,the driving factors of theme mutation are also explored to analyze the formation mechanism of mutation.The dynamic community evolution method is applied and improved to monitor the topic evolution and identify the topic mutation.The proposed method also helps discover the formation mechanism.This paper mainly includes the following four parts:First,this paper presents a community structure evolution tracking method to monitor the changes of community structure,including birth,expansion,disappearance,merging and splitting.At the same time,it can track the causes of critical changes in the community structure to explore the community structure evolution mechanism.Secondly,the critical changes of the community topological structure evolution are mapped to the topic mutations,to realize the real-time monitoring of the theme mutation.Then,based on the topic mutation monitoring,we explore the causes of the evolution and mutation of the theme,clarify the influencing factors and formation mechanism of the theme evolution and mutation,and realize the early warning and prediction of the theme mutation.Finally,the proposed method is applied to the keyword co occurrence network in the cloud computing field.The effectiveness of the method is verified by the experimental results.The main contents of this paper are as follows:(1)The method of tracking community structure of dynamic networks is applied and improved.Based on the node intimacy and degree,the weights of the links are recalculated.Then the algorithm of modular optimization is used to divide the static community structure.The algorithm is applied to the community partition of dynamic network according to the incremental clustering.Based on the short-term smoothness assumption,the historical community structure is considered to found the stable and persistent community structure.(2)Research on the community structure based mutation monitoring method.The sudden change of community structure in keyword co-occurrence network is used to represent the topic mutation,and then the identification and monitoring of theme mutation is realized.The method improves the accuracy,timeliness and result interpretability of topic mutation monitoring.