The Research And Application Of Link Prediction And Visualization In Networks Data

Big data is exploding as more and more information collected and stored daily.The way to fully mining the value of the data is a hot spot in academic circles.Currently machine is used to analyze those data.While we can't ignore that people play an important role in the data analysis.And in order to fully use the exploration and cognitive ability of human to discover the information in data,it's necessary to find the way visualizing the data information.If we can gain the predictive information of that unknown data and future data in the network and pass those predictions to users rapidly,it will provide reliable basis for users'decisions.What's more,users should be able to participate in the process of prediction by human-computer interaction rather than just observe the result data.In addition,as the continuously increasing of the network data scale,we need a filter or drilling method to make people more easily observe the data of interest,and obtain more effective information from the different size of network data.Therefore,the link prediction and visualization researches of network data have important practical significance.This thesis proposes an interactive prediction algorithm with the combination of complex network link prediction algorithm.It consults people's judgment to market more information in the form of human-computer interaction.This algorithm improves the prediction precision.In addition,this thesis proposes a visualization approach for the unknown data and the future data.Predict the unknown data and future data in the network.Then combine with the theory of complex network to visualization.Finally display the real-time prediction information.What's more,for the large-scale network data with ambiguous information and interaction problems,an iterative analysis and interactive exploration method is proposed.On one hand,it solves the problem of chaos and invalid information visualization when there are a large number of nodes.On the other hand,users can apply the interactive exploration method to drill down the network data,and change the network information to a suitable size.So that the data can be interacted and displayed more accurately.This thesis's interaction prediction method can get higher prediction accuracy results.And the data can be continuously explored.The prediction data visualization method in this thesis can help user understand the results rapidly and explore potential data information.Furthermore,the interactive analysis and exploration method for large-scale network data can adapt to all kinds of the scale of the network data.And according to the size of network data,this method can choose appropriate accuracy to appear more effective information.