Research Of Infor Mation Propagation Model Based On User's Dyna Mic Interaction Mechanism Via Social Network

Wit h the deve lopme nt o f computer a nd mob ile technolo gy,socia l network plat forms based on the Internet are more and more in-depth the pub lic's da ily life,work and study,ha ve become one o f the important places of people to spread and capture infor mat ion.Deep understanding o f user behavior and informat io n propagation mec hanis m on socia l network will be of great importance to pub lic opinion monitor and e-co mmerce market developme nt.However,the characterist ics of the rap idit y and nondeterminacy,and the comp lexity of propagat ion path caused by the cascading infor mat ion propagat ion based on user relat ions hips and other influence factors,have brought a great challe nge o f researching and modeling the user to propagate infor mat ion interact ive ly.In fact,the infor mat ion propagat ion on socia l network refers to the dyna mic interactive mecha nis m and the evo lut io n of network,user and infor mat ion,which need to build a model from mult i-leve l and mult i-sca le to achie ve the synergy ana lys is a nd mecha nis m understanding o f t he syste m in o rder.Exist ing research ma inly processing t he system fro m a single aspect of as the diffus io n result,unab le to effect ively solve the problem o f comp lexit y and nondeterminacy of information propagation.Therefore,this thesis combined the theory of infor mat ion science,comp lex network,and transmiss ion dyna mics,has proposed a n agent-based informat io n propagation model based on dyna mic interactio n amo ng users,by cons idering t he informat io n preferences and user relat io nships,in order to simulate and ana ly ze the informat io n spread on Enro n e-ma il service and Twitter socia l network service,respectively.The main work in this thesis includes:1)For the ema il infor mat ion communicatio n plat form,we built an ema il informat io n propagat ion model based on user 's dyna mic select ing behavior.First,we apply the Enron e ma il data and mine t he key properties of user and e mail,suc h as the user posit ion and posit io n collect ion of ema il partic ipants,etc,and then establis h the communicat ion rule between users and ema il infor mat ion preferences.On this basis,we propose an agent-based ema il propagat ion model by incorporating both the user influe nce of select ion of send ing to who m and e mail influence of select ion o f sending which.In part icular,the model uses a random-walk to simulate the frequency of user behavior.The exper imenta l results compared with the real data show that,the model could represent t he characterist ic distr ib ut ions and e merging patterns re lated to network,user and ema il.At last,we a lso study the impac t of interaction strengt h on propagation results.2)For the infor matio n propagation on Twitter soc ial network,we built a tweet informat io n propagation model based on the user's dyna mic retweeting behavior.First,we ana lyze the characterist ics of mult ip le c onnect ion on user follow network and mine different factors which influence the tweet propagatio n on interactio n network accord ing to real data.Further,based on analys is of dyna mic interact ive mecha nis m among network,user and tweet,we cons ider the user influence which is extracted fro m fo llow connect ion,the tweet influence whic h is measured as the cumulat ive number of tweet and the s imilar it y between tweet and user propagat ion history,and the user's limited attent ion for mass ive amounts of tweets,the n a agent-based tweet informat io n propagat ion model is proposed.The exper imenta l results s how an applicability and effect iveness of the model of explor ing the tweet propagat ion on Twitter social p lat form.At last,we also set an experiment to test the relat ions hip of initial tweet's releaser and its propagat ion maximization.The modeling and research of the infor mat ion propagat ion in this thes is based on user's dyna mic interactio n,is re lied on the basis of data mining and pattern recognit ion on soc ial network,and analyze quant itat ive ly the doub le coupling funct ion of mult id ime nsio na l re latio nship and infor mat ion preferences,and then establis h a dynamic e volut ion mode l.By co mparing the character ist ic distr ibut io ns and emer ging patterns fro m the perspectives of network,user and infor mat ion both on rea l data and model s imulat io n,we prove the applicabilit y and effect ive ness of the model for simulat ing the informat io n propagation on different scenar ios.Meanwhile,all the above studies provide a new so lut io n and met hod to solve the prob lem o f influe nce measure on infor mat ion propagatio n and va lid ity assessment of model,which has a certain sc ient ific significance and practica l significance of exp loring the impact of mult ip le factors on infor mat ion propagat ion and the network evolut ion fro m multi-level and multi-scale.