Note: To find revision information for really old versions of UCINET look in these two places:  Related Links
 To see a list of known bugs and user wish list items, check the fix list
 To download the latest version, click download.

News

Weeklong workshop on SNA
The LINKS Center at the University of Kentucky is offering its annual 1week summer workshop on social network analysis June 610, 2016 on the University of Kentucky campus ...
Posted Mar 15, 2016, 12:54 PM by Steve Borgatti

posted Oct 16, 2016, 5:29 AM by Steve Borgatti
[
updated Oct 16, 2016, 5:30 AM
]
 Added subgraph command to the CLI. Given a square matrix representing a network and a vector indicating membership of nodes in groups, the subgraph command allows you to extract a subgraph consisting only of nodes belonging to one or more groups. The syntax is <outfile> = subgraph(<input matrix> <attribute vector> <relational operator> <value>). For example:
>gender = cols(campattr 1) >boys = subgraph(campnet gender eq 1) >girls = subgraph(campnet gender eq 1)
If the operator and value are omitted, the program assumes "GT" and "0". For example, if you have a vector of 1s and 0s indicating who is a member of a given component, you could write:
>mainweb = subgraph(campnet maincomp)
This would select all nodes with a value greater than zero on maincomp.

posted Sep 8, 2016, 1:22 PM by Steve Borgatti
[
updated Sep 11, 2016, 11:46 AM
]
 Fixed discrepancy in how NetworksCompare DensitiesPaired and NetworksCompare DensitiesTheoretical parameter worked. Now, given the same random seed, they produce the same density standard errors

posted Sep 5, 2016, 3:19 PM by Steve Borgatti
 Added cosine similarity (aka Tucker's congruence coefficient) to ToolsSimilarities & Differences.
 Fixed bug in matrix editor which was adding rows and columns gratuitously

posted Jul 5, 2016, 7:56 AM by Steve Borgatti
[
updated Jul 5, 2016, 9:32 AM
]
 Added exact keyplayer routine to the CLI. The procedure finds the best set of k nodes which, if removed from the network, would maximize the extent of fragmentation (as measured by the number of pairs of nodes that can't reach each other by any path). The algorithm is brute force: it checks all possible subsets of size k. As a result, it is only appropriate for small networks and small k. For example, it can solve N=50 and K=3 in a few seconds, but N=100 and K=3 takes several minutes. Syntax is exhaustivekp(<dataset> <k>), such as:
>dsp exhaustivekp(zackar 3)
 Installer package contains version 2.160 of NetDraw, which changes the way the Width checkbox on the Rels tab works. Before, the width of the line between two nodes was determined by the average tie strength of all tie types currently active. Now, the width is determined by the largest tie strength of any currently active types of tie between that pair of nodes.

posted Jun 13, 2016, 11:11 AM by Steve Borgatti
[
updated Jun 13, 2016, 11:12 AM
]
 Fixed important bug in DataAffiliationsMode=Rows. When given a multimatrix dataset as input, it was writing out each 1mode projection twice.
 Replaced kplex routine in NetworkGroupsKPlexes

posted Jun 10, 2016, 4:49 AM by Steve Borgatti
[
updated Jun 10, 2016, 4:51 AM
]
 Made change in CLI to force use of courier new font. Previously, it was using the default Windows system font, which was almost always courier, but occasionally not. Using courier ensures that columns of numbers line up properly

posted Jun 5, 2016, 9:23 AM by Steve Borgatti
[
updated Jun 5, 2016, 9:24 AM
]
 Added new command DataFilter/extractExtract Columns. Use it to pull out one or more columns from a dataset and put them in a new dataset.

posted Jun 3, 2016, 3:40 PM by Steve Borgatti
 Fixed bug in ToolsTesting hypothesesNode levelRegression that was causing the program to always choose the first column of the dependent variable dataset
 Fixed bug in the matrix editor which was causing crashes on pressing the New button.

posted May 30, 2016, 3:14 PM by Steve Borgatti
[
updated Jun 1, 2016, 5:45 PM
]
 Fixed bug in the CLI's Forfiles command. Previously, commands of the form >forfiles samplk? draw were failing, while commands like >forfiles samplk* $cent = centrality($) were succeeding.
 Note: bug discovered in this version in ToolsTesting hypothesesNodelevelRegression. No matter which column you select as dependent variable, it always runs the regression on column 1

posted May 30, 2016, 7:07 AM by Steve Borgatti
[
updated Jun 1, 2016, 2:47 PM
]
 Note: bug discovered in this version in ToolsTesting hypothesesNodelevelRegression. No matter which column you select as dependent variable, it always runs the regression on column 1
 Added "number of holes" statistic to CLI's Holes routine. It simply counts the number of pair of alters that are not connected. This captures both network size and density
 Added limited Frequencies command to the CLI. Usage:
>dsp freq(sampson)
Matrix: Frequencies
1 2 3 4 5 6 7 8 9 10 SAM SAM SAM SAM SAM SAM SAM SAM SAM SAM PLK PLK PLK PDL PES PDE PIN PNI PPR PNP 1 2 3 K S N R           1 0 251 249 250 259 252 248 253 256 267 265 2 3 18 18 18 15 17 16 17 16 13 13 3 1 18 21 19 15 18 20 17 16 14 14 4 2 19 18 19 17 19 22 19 18 12 14
Matrix: Proportions
1 2 3 4 5 6 7 8 9 10 SAMPL SAMPL SAMPL SAMPD SAMPE SAMPD SAMPI SAMPN SAMPP SAMPN K1 K2 K3 LK S ES N IN R PR           1 0 0.820 0.814 0.817 0.846 0.824 0.810 0.827 0.837 0.873 0.866 2 3 0.059 0.059 0.059 0.049 0.056 0.052 0.056 0.052 0.042 0.042 3 1 0.059 0.069 0.062 0.049 0.059 0.065 0.056 0.052 0.046 0.046 4 2 0.062 0.059 0.062 0.056 0.062 0.072 0.062 0.059 0.039 0.046

