Note: To find revision information for really old versions of UCINET look in these two places:
- To see a list of known bugs and user wish list items, check the fix list
- To download the latest version, click download.
- Fixed bugs in the dialog box for Tools|Testing Hypotheses|Node-level|Regression.
- Removed in limitation in Network|Subgroups|Clique that was preventing the production of the clique-clique overlap matrix when the number of cliques was greater than 255. The new limit is 5000.
- Fixed bug in Network|Cohesion|E-I Index. It was screwing up the attribute vector when there were zeros present.
- Fixed bug in the DL editor. When importing data in matrix form in which there were duplicate labels (e.g., two nodes named "steve"), the program was incorrectly assigning all ties to the first of the duplicate nodes.
- 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.
- Fixed discrepancy in how Networks|Compare Densities|Paired and Networks|Compare Densities|Theoretical parameter worked. Now, given the same random seed, they produce the same density standard errors
- Added cosine similarity (aka Tucker's congruence coefficient) to Tools|Similarities & Differences.
- Fixed bug in matrix editor which was adding rows and columns gratuitously
- 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.
- Fixed important bug in Data|Affiliations|Mode=Rows. When given a multi-matrix dataset as input, it was writing out each 1-mode projection twice.
- Replaced k-plex routine in Network|Groups|K-Plexes
- 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