< Noteworthy Feature in NetMiner 3.4.0.d upgrade>

I. New Modules

1. Visualize >> Layout 3D, Drawing 3D

Layout and Drawing module that can be used for drawing large-scale 2D network map by separating the visualization process into two steps(layout and drawing) are now expanded to handle 3D network map. The coordinates of the 3D network map can be generated by Layout 3D module and drawing 3D network map using the generated coordinates can be accomplished by Drawing 3D module.

II. Enhanced Modules

Tremendous performance improvement is achieved in this update for some algorithms that required high time complexity.
According to the results of our performance test with large-scale network data(100,000 nodes and 1,000,000 links), the required time is improved from hours to ‘less than one minute’ for Transitivity, Ego-Network, K-Core module. Consequently, significant test method using Transitivity module in Triad Census and Triad Combination module is also improved. k-Plex module is also improved from few hours to a few minutes for the network data with 1,000 nodes and 10,000 links. For Clique and the n-Clique module, we added a new algorithm that results in performance improvement especially for scale-free and low clustering-coefficient networks.

< Noteworthy Feature in NetMiner 3.4.0.c upgrade>

I. New Modules

1. Visualize >> Layout, Drawing

Now the process for drawing network map can be separated into two steps using two new modules: 1) generating coordinates of the nodes on the network map through Layout module, 2) drawing network map based on the generated coordinates in the first step through Drawing module. These two modules can be very useful especially in the case of handling large-scale network data that requires much time complexity to draw network map.

II. Enhanced Modules

1. Transform >> Linkset >> Line Graph

- Options are added in Line Graph module to support more flexible and accurate data transformation and agorithm performance is improved to handle large-scale network data with low time complexity.

2. Statistics >> Autocoorelation

- Autocoorelation module which required too much time complexity to handle missing data is now improved.

3. Anlayze >> Centrality >> PageRank

- Algorithm performance of PageRank centrality moudle is improved to handle large-scale network data with low time complexity.

< Noteworthy Feature in NetMiner 3.4.0.b upgrade>

'Huge' License that can handle 1,000,000 nodes and 10,000,000 links is released.

<New Noteworthy Features in NetMiner 3.4.0.a upgrade>

I. New Modules

1. Transform >> Linkset >> Link Reduction
Link Reduction is a data transformation module for extracting links according to the weight of links with some options. Supported options are extracting method and handling tie value option.

2. Transform >> Matrix >> Vectorize
Vectorize (1-mode or 2-mode) module transforms a matrix into a vector. The node pairs or links in the input matrix become a nodeset in the transformed data. And the link weight will be transformed as an attribute of each nodeset.

3. Transform >> Mode >> Tree Construction
Tree Contruction is a data transform module for constructing tree type data from node attributes including multi-level hierarchical information. The constructed data can be added as a new 2-mode network data for related analysis modules (Analyze >> Position >> Expand / Collapse)

4. Analyze >> Neighbor >> Equicentrality
EquiCentrality is an analysis module that calculates the degree of similarity between the connected nodes’ centralities. Although it is similar to ‘Assortativity’ (Newman 2002), Equicentrality is based on the Euclidean distance. For users focusing on the distance of the node’s centralities, Equicentrality would be a good choice, other than ‘Assortativity’ which is based on correlation. (SoongMoon Kang, 2007)

5. Analyze >> Cohesion >> Community >> Label Propagation
Label Propagation for detecting the community structure of the network with a low complexity of O(m) is added as a new Community Detection Module. This is implemented simply by labeling all nodes, and then letting each node take the most frequent label observed from its neighbors, through iteration steps (U. N. Raghavan, R. Albert, and S. Kumara ,2007).

6. Analyze >> Position >> Expand / Collapse
Expand / Collapse is an analysis module for exploring hierarchical structure with flexible options. Blockmodeling results and related network map can be generated easily using flexible options for expanding and collapsing the level of hierarchical structure.

7. Analyze >> Property >> Modularity
Modularity is a measure to indicate how well the network is divided by the given partition. The NetMiner Modularity module simply provides an interface allowing the users to compute the modularity value for the network partitioned by specified vectors ( M. E. J. Newman, 2006).

II. Enhanced Features

1. Map Control Improvement
- Some useful user activities on the network map are supported as new features: deselecting pendant nodes, drawing network map with some selected nodes fixed, using mouse wheel in Visualize Neighbor sub window etc.

2. SVG Image Quality Improvement
- The quality of arrow color and node shape in SVG (Scalable Vector Graphic) image file is improved.

3. Link Threshold Control Item Improvement
Now users can select links on the network map more conveniently just by dragging slide bar according to the weight of links.

4. DB Connection Module Improvement
- Now MS Access can be selected as DBMS type for importing DB Tables into NetMiner (available on DB Connection Package).

5. Styling Option Improvement
- Users can select links on the network map by giving the minimum value and the maximum value. Since the function can be operated with a slide bar, it is more convenient for users.

6. Data Handling Option Improvement
- Output Tables can be opened in MS Excel more conveniently without taking trouble to copy and paste.

7. Decimal Point (Number Precision) Control Improvement
- Decimal place of the figures in the Output Results can be raised without limit.
- Decimal place can be set globally through more comprehensive Preference menu feature.

8. Automatic Update Support
- Now users can be alarmed about new updates on starting NetMiner.

III. System Enhancement and Bug-fix

1. Performance of P* (ERGM) model is enhanced. Now the algorithm runs faster and handles larger data than former versions.

2. The registration dialog is enhanced regarding off-line registration function. In fact, the function was added last year, but now the function can be accessed more easily.

3. Some of 'Out of Memory' errors are fixed.

<New Noteworthy Features in NetMiner 3.3.2 upgrade>

I. New Modules

1. Analyze >> Connection >> PFnet

PFNet module implements the PathFinder network algorithm that prunes the network so that the links do not violate the triangle inequality.

2. Statistics >> Logistic Regression

Logistic Regression Module is a model used for prediction of the probability of occurrence of an event by fitting train-set data to a logistic curve.

II. Enhanced Modules

1. Data Integrity Handling in Process Log

When modifying the input data on NetMiner3, previously all the sessions are discarded or new WorkFile is created. Now upon data modification, only sessions related to the modification are discarded.

2. Chart >> Matrix Diagram

- Color bar is newly available. Using it, users could check the color scale of the cells in the matrix diagram.
- The numerical values of each cell in the matrix diagram can be displayed, if users want.

3. Shortcut for editing data in Excel

- In data editor, it is possible to export data sheet to Excel program easily just through right-clicking of mouse.

4. Improvement of setting Preference feature

Now users can reduce repetitive option setting through more comprehensive Preference feature.

<New Noteworthy Features in NetMiner 3.3.1. upgrade>

I. Enhanced Modules

1. Analyze >> Connection >> Connectivity >> Node, Link

Performance of Node(Link) Connectivity Module is enhanced by algorithm improvement.

2. New feature of setting other Numerical Link Attribute as Link Weight

Weight of link can be substituted by other numerical link attribute in Editing Panel of Link List View using mouse right click menu ?Make Weight? in selected link attribute column.

3. Link Threshold Control Item in Select Panel

Link Threshold (Weight Threshold in prior version) Control Item is enhanced such that only links which meets the link threshold value for weight of link or numerical attribute data will be visible. Several threshold options are added: 1) Sorting Method (Descending, Ascending), 2) Extract Method (Value, Order), and 3) Handling tie value Options when ?Order? is selected as Extract Method. Missing value can be handled by ?exclude? and ?include? options as well. Current Process can be run again with only the selected links on the network map using ?Run with showing links? button.

4. Recursive Use of Output Result of 1-mode Network type as Link Attribute

Output Result of 1-mode Network type can be added as link attribute of 1-mode(2-mode) network data in Output Panel using mouse right click menu "Add to Workfile >> 1-mode Link Attribute".

5. Styling Multiple Node Attribute in a single dialogue box

Node styling properties such as Shape, Surface, Size, Label, etc. can be chosen in a single Node Style dialog box.

II. Fixed Bug reports

Several bugs were fixed in 1) Dichotomize and Reorder modules of Transform Menu, 2) Dyad Census and Lambda Set of Analyze Menu, and 3) visualization.

<New Noteworthy Features in NetMiner 3.3.0.a upgrade>

I. Enhanced Modules

1. Multiple Regression QAP (MRQAP)
Post-processing option for "Statistics >> Regression >> Matrix" module is enhanced with Multiple Regression QAP significance test. There are five options for users to choose - X-Permutation, Y-Permutation, FLSP, DSL, Partialing.

2. Standardization option in Regression modules
Before running regression analysis, users are able to standardize their data. The standardization option is at Pre-processing section of regression modules..

II. Modification of Names and Locations

1. Visualize >> Clustered
Default layout algorithm of clustered map has been changed from Clustered-CoLa to Clustered Eades.

III. Fixed Bugs

1. Analyze >> Position >> Brokerage
Previously if data contained an underscored character, it used to be mistakenly displayed as a subscript in the input control item.

2. Statistics >> Autocorrelation >> Join-Count
This module did not work if there was at least one node whose attribute value was missing.

3. Transform >> Query
A crash reported from selecting '@More' in the value-list box of the string type data of the query module if there were missing string-typed values at "Transform >> Query" module.

<Hot Features in NetMiner 3.3 Update>

I. New Modules are added to NetMiner 3.

1. New network measure ‘Assortativity’ is added. (Package: Network Measure-Advanced)

2. Now NetMiner provides the most probable estimation of power-law model. (Package: Statistics-Multivariate Analysis)

3. Spring-embedding layouts are extended to 2-mode Networks. (Package: Network Visualization – 2D)

4. New charting module of 'Pie Chart' and 'Box Plot' has been added. (Chart - 2D)

5. ‘Transform >> Query’ module is added to NetMiner 3. It substitutes for both ‘Node Extract’ and ‘Link Extract’ module. (Package: What-if Analysis)

II. Processing modules are enhanced.

1. Some modules have new outputs. ex) 'Centrality' measures, 'Multiplex', 'Network Properties', 'Brokerage', 'Cohesive Block' and etc.

2. Options are enhanced. Some modules have new options, and some modules' options become more elaborate: 'Equivalence', 'Degree', 'Merge', 'Homophily' and etc.

3. Default Style is applied to maps. For example, the nodes' size of Spring Map (which is created by 'Degree' analysis) is proportional to the degree score of each node.

4. Interface of recursive data use is more user-friendly now. For instance if a data is transformed, dialog window pops out. With this dialog user's able to add the result directly.

5. New items are added in Inspect Panel, called 'Explore' Panel in previous version, of some modules such as 'Node Connectivity', 'Shortest Path', 'Structural Hole' and etc.

III. New Features

1. Auto Save and Auto Recovery: Now NetMiner 3 saves your data automatically. If the software is shut down unexpectedly, it shows auto-saved data at next software execution.

2. Easy project-handling: Handling project is more user-friendly in 3.3 version. For example, project-creating procedure became simple. And one project is saved as one NMF file.

3. Enhanced importing/exporting tool: Importing and exporting are enhanced in NetMiner version 3.3. GML and StOCNET files can be imported to NetMiner, and interface of importing/exporting tools has been improved to be more user-friendly.

4. Enhanced General User Interface: There are changes in icon shape in Toolbar, Current Workfile, and Workfile Tree.

5. Enhanced Editor

6. Context-sensitive Data Management: New options are added in Data Context Menu (right click in 'Current Workfile' and 'Workfile Tree'.)

7. More interactive and easier Visualization: For example, a user is able to save selected nodes as a new Workfile.

8. Enhanced Random Data Generation: User interface of Random Data Generation is enhanced, and new data generating algorithms are added.

-If you are interested in finding out 3.3 updates in detail, please click here.

(New Feature in previous packages)

[Network Measures - 2-mode]
Degree analysis is extended to 2-mode network in the menu of 'Analysis > Two-mode > Degree'.

[Network Visualization - 2D]
New layered drawing algorithm 'Dig-CoLa' is added. This algorithm is similar to Kamada-Kawaii style(thus visualize connectivity) in terms of layout, while nodes are layered to show the overall "flow" of the graph.

[Network Visualization - 2D]
New clustered graph layout algorithm 'Clustered-CoLa' is added. This algorithm separates nodes of different clusters very clearly, while graph style will look similar to Kamada-Kawaii style.

[Network Visualization - 2D]
New graph layout algorithm 'Stress Majorization' is added. The output of this algorithm is exactly same to that of Kamada-Kawaii algorithm. However, this algorithm is monotonically converging so that you would not worry about non-convergent behaviour of Kamada-Kawaii algorithm. Therefore, it is very useful when you refine your layout.

(Major Improvements)

- You can now sort/filter result tables in context menu, which is shown when you right-click on it. You can both sort in increasing/decreasing order, and excel-style filtering can be done on node attributes.

- If you select nodes/links in the graph map and right click those, you can save shapes of nodes/links in the context menu. Once the shape of node/link is saved, it is chosen as a default shape of node/link, when you create a new visualization session.

- Whenever you press 'F1', context-sensitive menu pops out.

(Minor Improvements)

- More smoothly you can zoom in/out with wheel button on the graph map.

- 'Save as new Workfile' function of 'Transform > Proximity > Affiliation' now provides user the freedom of selection between Overlap network and Bipartite network.

(New Package)

[Matrix Calculator]

Now NetMiner provides command-line interface (just like that of MATLAB Command-Window or Scientific Calculator) that enables user to effectively manipulate matrices. Users with knowledge in matrix mathematics can more flexibly do the analysis within the framework of NetMiner.

<New Feature in previous packages>

[Measures Advanced]

New community module of 'Cohesion > Community > Eigenvector' is added.
It uses leading eigenvector of modularity matrix to find communities.
This method is known to provide best in modularity of its result compared to other algorithms.

[Measures Advanced]

Community Centrality is introduced in new module 'Centrality > Community'.
Community Centrality measures how central the node is in its local neighborhood.

(Major Improvements)

- Now Node/Link Coloring Wizard in Graph Map can be quickly run with Icons on Toolbar or Shortcut Keys (Ctrl-Shift-N/L)
- Now you can quickly layout the Graph Map with icon on Toolbar or Shortcut Key (Ctrl - L)
- You can filter nodes with those you selected in visualized Graph Map by right clicking on map and selecting 'Run with Selected Nodes' in context menu. You can roll back with 'Run with Entire Nodes'

(Minor Improvements)
- You can zoom in/out on Map with mouse wheel.
- When exporting Node Attributes in Excel file, the cell type is automatically determined.
- Isolated nodes in Graph Maps can now be manipulated more gracefully.

(Bug Fix)
- 'Windows > Project View' used to behave abnormaly.
- 'Cohesion > Community > Agglmorative' used to behave abnormaly, when the graph was disconnected.
- In 'Import Datafiles Wizard', the contents of an excel file used not to be not updated after reading once.
- When importing node attributes from csv/xls file, it used to stop reading a row after finding a single missing value without reading to the end of line.

(new feature)

[DB Connection]

Users can now retreive network data from Oracle or SQL Server(MS SQL).

[Visualization]

New layout named 'HDE' has been introduced, which is super-fast even for large size of data.

(major improvement)

[Graph Editor]

'Graph Editor' has been integrated into main program.
Adjacency Editor now provides three kinds of view, 'Matrix', 'Edge List' and 'Graph Editor,' each of which is different presentation of same data. Therefore, it is much easier and more intuitive to edit your data.

[Cohesion > Community]

Now 'Community' is divided into two menu items, 'Betweenness' and 'Modularity'.
'Betweenness' provides 'GN' algorithm, which is based on betweenness analysis.
'Modularity' provides popular 'CNM(Clauset, Newman and Moore)' algorithm and its variants, which is highly applicable to large networks.

(minor improvement)

[Visualization]

Component arrangment method of 'Tiling' has been improved.
Now it generates layouts with proper margins and alignments.

The performance of revised 'KK' Algorithm has been improved to run much faster.

(bug fix)

[Visualization]

The result of 'KK' algorithm was unstable when it was run iteratively with
'Initial Point' option of 'Current Position'. Now it is fixed to give stable result.

[Data Transformation]

Function modified

• Ego-net Extraction, including Alter-to-Alter links could be optional
  : Transform>Node set>Ego-net

[Map View Functionality]

Function modified

• Coloring
  : Coloring status remains after closing option dialog.
  (Using View>coloring>Default, users can restore a default map.)

[Network Analysis]

New functions added

• Marsden's role typology
  : Analysis>Neighbor>Role Typology
• Newman's fast Community detection
  : Analysis>Cohesion>Community(Fast)
  

[Network Visualization]

New functions added

• Kruskal's MDS
  : Visualize>MDS>n-MDS(Kruskal)
• Scatter Plot
  : Visualize>Chart>Scatterplot

[Statistics]

New functions added

• Kruskal's MDS
  : Statistics>MDS>n-MDS(Kruskal)
• Network Contingency Table
  : Statistics>Crosstabs>Network
• Covariance Matrix
  : Statistics>Covariance Matrix
• Principal Component Analysis
  : Statistics>Principal Component
• Factor Analysis
  : Statistics>Factor Analysis
• Gini Coefficient
  : Statistics>Gini Coefficient>Vector/Matrix

[Menu Structure Reorganization]

• Analyze>Cohesion>Community --> Analyze>Cohesion>Community(GN)
• Visualize>MDS>n-MDS --> Visualize>MDS>n-MDS(Alscal)
• Visualize>MDS>n-MDS(3D) --> Visualize>MDS>n-MDS(Alscal)(3D)
• Statistics>MDS>Classical MDS --> Statistics>MDS>c-MDS
• Statistics>MDS>Non-metric MDS --> Statistics>MDS>n-MDS(Alscal)

• Import and Export list-type of network data (File | Import [Export] | CSV | List menu): support Adjacency or Affiliation data which labeled with character-string
• 'Data open dialog' shows when a data file is loaded

  2. Analysis menu

• Subgraph: Dyadic Interaction(p1) - Group Dyadic Interation(p1) model was integrated to existing routine
• Centrality: Random-walk Betweenness Centrality (originated by M.E. Newman), Coreness (k-Core number originated by Vladimir Batagelj)
• Equivalence: SimRank routine was modified to input dampening parameter
• Position: Generalized Blockmodeling was newly added (originated by Vladimir Batagelj)

  3. Statistics menu

• Autocorrelation: Assortativity was newly added (Assortative mixing orginated by M.E. Newman)