SONIVIS User Manual v0.8

From SONIVIS:Wiki

Jump to: navigation, search
This article is work in progress.



The following page provides an overview of using SONIVIS:Tool. This tool allows users to load, analyze, and visualize social and information networks computed from MediaWiki based Wiki.

Contents

Concept

SONIVIS:Tool is a Java-based open source software, which is based on [Eclipse Rich Client Platform] (RCP). Eclipse was developed proprietarily by IBM and published under an open source license in 2001. RCP enables the creation of generic applications with a native graphical user interface. It offers a simple dynamic component-based model to develop plug-ins that contains the functionality of the application. The metrics are calculated by [GNU R], an open source software environment for statistical computing and graphics. SONIVIS:Tool essentially follows the model-view-controller (MVC) pattern, that means data, functionality, and user interface are separated.

To enhance the usability of SONIVIS, two different perspectives are offered: the analysis and the manipulation perspective. Each perspective contains different views, which are predefined in their position and size. Following the Eclipse approach, these views can be organized based on the user's needs.

Download and install the recent version of SONIVIS

The current version of SONIVIS is for Windows here and for Linux here available. In the binary version we offer you an executable version which sets all necessary system variables automatically.

If you use Windows, we recommend to install directly SONIVIS on c:\.

First start of SONIVIS:Tool: Setting up the tool

Database Connection for version 0.8

During the first run of the tool various settings must be done. It is also possible to change the settings later via the Window->Preferences menu item. First, you have to select the measures that you need for your analysis. To begin an easy analysis, select Betweenness Centrality (This measure shows the influence of a node on its network by measuring how many paths through the network go through a node), Degree (Number of neighbors of a node, e.g. number of co-authors of an author in a collaboration network), Amount Contribution (This shows how many bytes an author has contributed to the information space) and Eigenvector Centrality (connectedness of a node to the core of the network). If you need more information about measures, you find a description of all measures in the metric portal.

To allow the tool to load its data from the database, you also have to configure the database connection to your MySQL server. After this, the most important settings are done and you can start your analysis session. (Help for installing the database can be found in the Installation manual).

Help for working with SONIVIS:Tool can be found in the User Manual.

General

How to open a perspective?

Selection of perspective

There are two perspectives in the open source version and three perspectives in the binary version available. After the first start of SONIVIS, only one perspective - the analysis perspective - is visible. The other views have to be loaded for the first time.

  1. Choose "Window" in the main menu.
  2. Choose "Open Perspective"
  3. Click on the desired perspective.
  4. You can rearrange the sequence of all perspectives using drag-and-drop.

It is also possible to define your own perspective based on your requirements. The first step is to organize the different views as you want. Then save this new defined perspective with "Save Perspective As...". This new perspective is still available after a re-start of SONIVIS.

How to open a view?

Selection of view

A view is typically used to show specific information, provide selection boxes, or display properties of another active view. The position of each view is freely definable. There are only few restrictions on changing the layout of a specific view. Therefore, the usage of predefined perspectives is not obligatory. Users can change the layout of the pane depending on their specific requirements.


  1. Choose "Window" in the main menu.
  2. Choose "Show View"
  3. Click on the desired view.
  4. After the view has opened you can define its position per drag and drop.


A view can be closed or reduced in size. If you want to do this please use the buttons on the top right of each view.

Configuration

How to open the preferences?

How to configure a database connection?

Network

How to load a network?

Before beginning an analysis, you should activate only the measures that you need in the Metrics Preferences dialog. Additionally, you should limit your analyses by filtering the network by category or namespace. This can save you a lot of time.

Here is a step-by-step guide how to load a network:

  1. Select a network type (mandatory).
    At the moment, there are three network types available: collaboration network, wiki-link network, and dynamic collaboration network.
  2. Specify the start and end day of your network visualization (optional).
    If there is no date chosen, the starting day is 01.01.1970 and the ending day is today.
  3. Select a namespace (optional).
  4. Select a specific category (optional – see note above).
  5. Click on the "Load Network" button.
    If a category is selected, the network is loaded by using the return key.

After loading (here you have to setup a connection to the database, see description here) the selected databases, basic domain specific measurements are conducted automatically and visualized in the characteristics view (number 1 in picture). These metrics are categorized in general domain measures and attribute measures. There is already a specification for MediaWiki.

In addition, first statistical measurements and temporal measurements are visualized graphically in the statistics view (number 1 in picture). These diagrams offer a quick overview of the existing developments in the investigated information space.

How to zoom in or out in a network?

Right-click on the network graph and drag the mouse to zoom in or out of the network graph.


How to (de)select a node?

There are two possibilities of select a node. One possibility is as follows:

  1. Load a network.
  2. Click on the desired node in the network. This node and all adjacent edges are highlighted.
    To unselect a node keep the control (CTRL) key pressed and click on that node again.
    You can select more the one node. During the selection process you have to keep the control (CTRL) key pressed.

You can now use the "Hide node" filter in the Network Filter view to remove the nodes visually from the network.

Measures

How to activate measures?

How to load/display measures?

How to show node measures?

  1. Define a database connection
  2. Load a network
  3. Choose the specific node in the network.
  4. The node characteristics are shown on the right hand side in the node properties view.
    Based on your selection in the Preference pages, all calculated metrics are shown.

When the network data are non-anonymous, node names are displayed as the node IDs. Furthermore the weight of the relations between the nodes is shown depending on the selected network. Simultaneously to a visualization of the network, specific measures on the macroscopic level are calculated and represented. The same applies to node measures that describe the microscopic level.



How to export measures?

Statistics

How to activate predefined statistics?

Preference Page - Statistics

Predefined statistics help to identify changes in the investigated domain quickly.

  1. Choose "Window" in the main menu.
  2. Select "Preferences..."
    The preference pane is organized in two parts. On the left hand side is the topic tree and the right hand side depends on your selection.
  3. Click on "Statistics" and set the check mark on "Activate Statistics".
    Each created image will be saved in the folder you enter in the "Select Folder" text field. By default, images are saved in the SONIVIS folder.


Clustering

1. Load a network.

2. Choose DataMining -> Clustering.

3. Select in the following wizard your options:

Method:
1. partioning
2.hierarchical
Algorithm:
1. partioning has the following algorithms to choose from:
k-means
PAM
1. hierarchical has the following algorithms to choose from:
single linkage
complete linkage
average linkage
WARD
Parameter:
1. select one or more of the listed metrics you wish to cluster
Amount of clusters:
1. select the number of clusters you want to calculate

4. Press Finish to calculate the clusters.

The calculated clusters are shown as colored hulls in the GraphView and you also get a list of all clusters in the CLusterView.

Personal tools