Net Explorer

Interactive network exploration widget.

Channels

Inputs

Network (Network)
Network object to plot.
Items (ExampleTable)
Additional data about vertices, one example for each vertex.
Items to Mark (ExampleTable)
A subset from items instances to mark.
Items Subset (ExampleTable)
A subset from items instances to show. All other instances will be hidden.
Vertex Distance (SymMatrix)
Distance matrix with distances between vertices.

Outputs

Selected Network (Network)
Subnetwork of all selected nodes.
Selected Examples (ExampleTable)
Descriptions of vertices (from the graph of input signal Items) that correspond to selected vertices.
Unselected Examples (ExampleTable)
Descriptions of vertices corresponding to unselected vertices.
Marked Examples (ExampleTable)
Descriptions of vertices corresponding to marked vertices.

Description

Net Explorer visualizes graphs, lets the user explore them by selecting their individual parts and gives the corresponding data to other widgets. It is currently available only in the latest (Qt4) version of Orange.

More information on the use the widget from Python is available in documentation on orngNetwork Orange module.

Classification Tree Graph
- Node Tab

Navigation

Four navigation buttons are located in the upper right corner of the control area.

Zooming:

Zooming mode is selected by default. User can zoom in a rectangle by pressing and holding the left mouse button and dragging the mouse to draw a rectangle to zoom in to. Zoom out (one level) command can then be executed by pressing right mouse button.

Zoom to extent:

By pressing this button the zoom is automatically set to show all visible nodes in the network.

Zoom selection:

By pressing this button the zoom is automatically set to show all selected nodes in the network.

Panning:

Use this mode to drag the network around the canvas.

Marking and Selecting

The widget features a two-stage procedure for selection of vertices, which allows for a very flexible manipulation of subsets of vertices. Vertices can be marked and/or selected. Marked vertices are presented by filled circled and selected vertices are highlighted with a yellow border.

Vertices can be selected manually by clicking or by drawing selection rectangle. Selection mode is enabled by pressing fifth button from the top in the upper right corner in control area.

Another way to select vertices is to add or remove the marked vertices to or from the selection, or to replace the current selection with the marked vertices (buttons seven to nine).

The vertices can be marked based upon values of the attributes of the corresponding objects. Most marking options are available in the Mark Tab.

Vertices Tab

Parameters in vertices tab govern the appearance of network vertices and data associated with them. User can select optimization method, vertex color and size attribute, set the labels and tooltips and other rendering options.

Optimization methods

No optimization
The vertices are left as they are. If the given network object already contains the vertices placement data (these can be given, for instance, in Pajek format), they are placed accordingly, otherwise their positions are random.
Random
The vertices are scattered randomly.
Fruchterman-Reingold
The standard Fruchterman-Reingold algorithm, which tries to put the pairs of connected vertices to a certain fixed small distance and the unconnected ones to the fixed large distance. A simulated annealing optimization algorithm is used to minimize the stress of the solution.
Fruchterman-Reingold Weighted
A variation of the Fruchterman-Reingold algorithm which also takes the edge weights into account: the larger the weight, the smaller the desired distance between the two vertices.
Fruchterman-Reingold Radial
A Fruchterman-Reingold-type algorithm which places a vertex selected by the user at the center of the graph and optimizes the layout around it. The optimization procedure ensures that vertices with shorter paths to the central vertex are closer to it than those with longer paths.
Circular
The vertices are placed around the circle in the same order in which they are given in the data.
Circular crossing reduction
A local optimization algorithm which puts the vertices on the circle and tries to minimize the number of crossed edges.
Circular random
The vertices are placed randomly around the circle.

The user can specify the number of iterations of the optimization procedure where applicable.

Labels and tooltips

One or more attributes can be chosen to be shown as Vertex labels or tooltips. To reduce the visual clutter, the user can also decide to Show labels on marked vertices only. Checking Inside view enables a special mode in which only selected vertex and vertices closer to it than the specified distance (in terms of the number of connections, not their lengths) are shown.

Show missing values decides whether missing values should be displayed (as question marks).

Edges Tab

Parameters in edges tab are used to specify the appearance of edges. One can set edges color, width and whether weights of edges are displayed.

When Explore vertex distances is selected, the user can examine the distance between vertices, as given in the "Vertex Distance" signal. To display the distance, one must first select a vertex. When hovering over an arbitrary vertex, the distance between the selected vertex and the vertex under the mouse pointer is displayed as a popup.

Mark Tab

Most options for interactive network exploration are gathered here. User can search for nodes by value in items meta data, mark neighbours of selected vertices or mark vertices by number of connections.

Info Tab

The info tab gives some general information (such as number of vertices, number of edges, diameter, clustering coefficient) about the network. Button Degree Distribution calculates and displays the degree distribution. Finally, the network can be saved in Pajek format by clicking on Save network button.

Examples

The best way to get to know how to use this widget is by playing with it. Try some of the following network data sets:

More comprehensive list of examples will be available soon.