Research Of User Influence Algorithm Based On Improved Linear Threshold Model And Platform Implementation

With the development of economy and the continuous progress of network technology,micro-blog has gradually become an indispensable part of people's daily life.We know that through tweeting,forward,like,do comment to other micro-blogs,the relationships contact with other people around becomes increasingly close.At the same time,the research on micro-blogs and micro-blog users are gradually increasing popular.Through the studying of micro-blog users we found that some users are more popular than others in the virtual world of micro-blog.We call these people "opinion leader".Apparently,these opinion leaders plays an important role in the micro-blog system and their every word and action also has a great influence on the whole network.For these opinion leaders,companies can promote their products through them;the media can spread the news through them more rapidly.But every coin has two sides,it also makes some criminals through them in the blog system.With the large amount of data,the high velocity it spread news the micro-blog transfer the real life events to the virtual world.Anyhow,whatever point of view,as the social network relationship has become more and more complex,the studying of social network is become more and more imminent.Social networks have a large number of complex relationships,to the studying of social networks,we can either start from the user,or from micro-blog to expand.In this paper,based on the relevant work,we propose an improved linear threshold model.And based on the existing micro-blog system we design API to facilitate the use of third party platform.The main research contents and results of this paper include:1)From the network structure,we analysis the node attribute of network.According to the background of the network information,we run the most forward relation,the average distance to the network and forwarding rate model and conduct experiments of the three models.2)To the characteristics of micro-blog users,we analyze the user's active degree from the interactive information of micro-blog and other bloggers in the blog text.Then,based on the traditional linear threshold model,we add the user's activity to the threshold of users to improve the linear threshold model.Finally,based on the real data set,the results of improved linear threshold model is verified by experiments.Finally,we get the approximate degree of 63% to get the TOP-K users.3)In addition,due to the research of existing user influence analysis lacking of corresponding results displaying platform,in order to release the researchers from the complex development of the foreground development and background development of platform,also for the ongoing of the research of information dissemination and the presents the results of our research,we developed the user influence analysis platform API and user behavior simulation function with making an appropriate use of technologies of CXF and Axis2.We can simulate user's behavior in the micro-blog through the simulation function,like tweeting,forwarding,liking and commenting;we can access and modify background data without visiting the frontend of the website through the API in the way of remote procedure calling.For the API,in order to facilitate our using,we developed the WSDL document for each API with the detailed instructions with WSDL format.Finally,we test the APIs and simulation functions,and the results show that we can satisfy the required functions.4)Seeing from the perspective of the system,we design and implement the platform system based on influence algorithm.In order to show the results of our research,we have developed the platform framework consisting of the function from the data acquisition to the micro-blog system to users' influence ranking.In order to dynamically update the results,we have joined dynamic updating mechanism to the influence ranking module.