How could I use graph mining method to get a multi-node graph?

Question

I now use apriori algorithm to do a data mining project,and I get result such as:item1 <=> iteam2、item2 <=> item3....... I want use graph mining to generate a graph containing many nodes and illustrating relation between these node like this：

I heard some data ming software--weka,rapidminer;I also heard some graph library--igraph,networkx;I also heard--tableau.But I'm still confused,could someone give me a illustration about detailed procedure?

Answer 1

I recommend using Prefuse tool kit for your problem. Take a look here http://prefuse.org/gallery/ . This contains an example of the graph that you need.

Loosely speaking, Prefuse also has a browser version called D3.js . If you want to display your graph in browser then use D3.js

I have used Prefuse as well as D3.js when I needed a desktop graph and a graph in the browser.

Answer 2

If by multi-node graph you are referring to this definition http://dl.acm.org/citation.cfm?id=1292799, then I would say that you could use Gephi to visualize your graph. Gephi is a powerful tool for network visualization/analysis, since you can annotate the vertices, apply clustering algorithms etc.

In your case, since multi-node graphs have multiple states, you can either use some annotation/coloring to show the different states of the nodes/edges, or even visualize these different states by importing different timestamps/versions of the network in Gephi. You can then observe the differences among them. Even if your graph is not multi-node, I would recommend Gephi for visualizing it.

If item1 <=> iteam2、item2 <=> item3 .. is your current data format, you can transform it to a format that Gephi recognizes, like adjacency list or edge list.