Remerciez-le!

Remerciez @Admin pour avoir partagé cet document gratuitement, de la manière la plus simple, en partageant sur les réseaux sociaux.

UCINET 6 for Windows Software for Social Network Analysis USER'S GUIDE Borgatti

UCINET 6 for Windows Software for Social Network Analysis USER'S GUIDE Borgatti, Everett and Freeman 2002 Copyright © 1999-2002 Analytic Technologies, Inc. Last revised: 31 January, 2015 UCINET 6  1999-2002 Analytic Technologies. All rights reserved. Analytic Technologies 11 Ohlin Lane Harvard, MA 01451 USA Voice: (978) 456-7372 Fax: (978) 456-7373 Email: support@analytictech.com; sales@analytictech.com Table of Contents Preface 0 Notational Conventions 0.1 Acknowledgements 0.2 Programming Considerations 0.3 Content 0.4 Matrix Orientation 0.5 Getting Started 1 Hardware 1.1 Installation 1.2 Quitting the Program 1.3 Technical Support 1.4 Citing The Program 1.5 License and Limited Warranty 1.6 The UCINET Environment 2 Menus and Help 2.1 Forms 2.2 Running An Analysis 2.3 The Log File 2.4 Datasets 2.5 Program Organization 2.6 File Submenu 2.7 Data Submenu 2.8 Transform Submenu 2.9 Tools Submenu 2.10 Networks Submenu 2.11 Options Submenu 2.12 Importing Data 3 RAW Filetype 3.1 Excel Filetype 3.2 DL Filetype 3.3 Full Matrix Format 3.4 Rectangular Matrices 3.5 Labels 3.6 Multiple Matrices 3.7 External Data File 3.8 Diagonal Absent 3.9 Lowerhalf and Upperhalf Matrices 3.10 Blockmatrix Format 3.11 Linked List Formats 3.12 Nodelists 3.12.1 Edgelist Formats 3.12.2 Edgearray Format 3.12.3 UCINET Spreadsheet Editor 3.15 Data Processing 4 Subgraphs and Submatrices 4.1 Merging Datasets 4.2 Permutations and Sorts 4.3 Transposing and Reshaping 4.4 Recodes 4.5 Linear Transformations 4.6 Symmetrizing 4.7 Geodesic Distances and Reachability 4.8 Aggregation 4.9 Normalizing and Standardizing 4.10 Mode Changes 4.11 Where is it now? 5 ~ 0 ~ Preface 0.1 Notational Conventions UCINET is menu-driven Windows program. This means you choose what you want to do by selecting items from a menu. Menus may be nested, so that choosing an item from a menu may call up a submenu with additional choices, which in turn may call up submenus of their own. Consequently, to get to certain choices, you may have to select through a number of menus along the way. To represent the options you must take to a given choice, we use angle brackets. For example, to run the hierarchical clustering procedure, you must first start UCINET, then click on the top toolbar and point to Tools, from the drop down menu that appears highlight Cluster and from the submenu that appears click on Hierarchical. We will represent this series of choices as Tools>Cluster>Hierarchical 0.2 Acknowledgements Dozens of people have contributed to UCINET 6 for Windows by making suggestions, contributing technical expertise, finding bugs, and providing moral and financial support. We especially thank Charles Kadushin, David Krackhardt, Ron Rice and Lee Sailer for their early and continued support. We also thank those who have contributed technical expertise for specific algorithms, including Pip Pattison (semigroups), Kim Romney (correspondence analysis) and Stan Wasserman (p1). We are also grateful to Brian Kneller (University of Greenwich, London) for doing much of the initial programming on the Windows interface. The non-metric MDS program is adapted from UCINET 3.0 (MacEvoy and Freeman, 1985), which in turn was drawn from the University of Edinburgh's MINISSA program. Many of the procedures make use of routines found in Numerical Recipes in Pascal by Press, Flannery, Teukolsky and Vetterling, and EISPACK (Smith et. al. 1976, Springer-Verlag). 0.3 Programming Considerations To paraphrase an old song (and reverse the meaning), UCINET 6.0 is built for speed, not for comfort. Oftentimes during the programming of UCINET, we had to choose between using a fast algorithm that used a lot of memory (and therefore reduced the maximum size of network it could handle), and a slow algorithm that saved memory and could handle much larger datasets. In previous versions we tried to strike a balance between the two. In this version, we usually chose speed. One reason for this is that it is precisely when working with large datasets that speed is essential: what good is an algorithm that can handle thousands of nodes if it takes days to execute? The other reason is that advances in hardware and operating system software continually extend the amount of memory programs can access, so it seems a waste of programming time to work out ways to economize on memory. One consequence of menu systems is the need to organize program capabilities into categories and subcategories in a way that is logical and comprehensible. Of course, this has proved to be impossible. With only a little effort one can discover several competing schemes for classifying all the functions that UCINET 6.0 offers. None of the schemes is perfect: each does an elegant job of classifying certain items but seems forced and arbitrary with others. The scheme we have settled on is no exception. The basic idea is that under "network" we put techniques whose reason for being is fundamentally network-theoretic: techniques whose interpretation is forced when applied to non-network data. An example of such a technique is a centrality measure. In contrast, under "tools" we put techniques that are frequently used by network analysts, but are also commonly used in contexts having nothing to do with networks. Multidimensional scaling and cluster analysis are examples of such procedures. Inevitably, of course, there are techniques that are either difficult to classify or for some reason are convenient to misclassify. 0.4 Content UCINET 6.0 is basically a re-engineered version of UCINET IV, and so users familiar with UCINET IV should easily adapt to the new environment. We have extended UCINET's capabilities and re-organised the routines into what we believe to be more sensible categories. Perhaps the most fundamental design consideration we have faced is choosing what capabilities the program should have. Since Freeman's first version was released, UCINET has incorporated a diverse collection of network techniques. The techniques are diverse both in the sense of what they do (detect cohesive subgroups, measure centrality, etc), and where they come from (having been developed by different individuals from different mathematical, methodological, and substantive points of view). In UCINET 6.0 we continue that tradition, seeing ourselves more as editors and publishers of diverse works than as authors with a single pervading perspective. One problem with this approach is that different techniques implicitly assume different views of what their data are. For example, graph-theoretic techniques describe their data in terms of abstract collections of points and lines, algebraic techniques view their data as sets and relations, and statistical techniques understand their data to be variables, vectors, or matrices. Sometimes the mathematical traditions intersect and the same operation is found in the repertoire of each tradition; cognates, if you will. For example, a graph automorphism, which is a 1-1 mapping of a graph to itself, corresponds in the world of 1-mode matrices to a re-ordering of the rows and corresponding columns of a matrix. Likewise, the converse of a graph, in which the direction of all directed arcs is reversed, translates to the transpose of a matrix, which is an exchange of rows and columns. Unfortunately, there are also false cognates among the lexicons. For example, a relation may be expressed as a matrix, but in discrete algebra, the "inverse" of a relation is the transpose of that matrix, not the matrix which, when multiplied by the original, yields the identity. 0.5 Matrix Orientation In UCINET 6.0, all data are described as matrices. While the prompts and outputs of some procedures may reflect the language most commonly associated with that technique (e.g. "networks" for centrality measures, "relations" for ego-algebras), it is extremely important for the user to maintain a "matrix-centered" view of the data. All UCINET data are ultimately stored and described as collections of matrices. Understanding how graphs, networks, relations, hypergraphs, and all the other entities of network analysis are represented as matrices is essential to efficient, trouble-free usage of the system. ~ 1 ~ Getting Started 1.1 Hardware UCINET 6.0 requires a computer running Windows 95 (from 1997 or more recent), Windows 98, NT or other compatible operating system. The program requires 2mb of hard disk storage space and 16mb of RAM. 1.2 Installation The UCINET program must be installed before it can be used. If you have a CD then place the disk in the drive and the program should start the installation automatically. If this does not happen then from the start button select the run option change to your CD drive and click on "setup.exe". If you have an electronic version then from run select the folder containing UCINET and select the file "setup.exe". The installation wizard will guide you through the installation procedure. 1.3 Quitting the Program To leave the program, choose File>Exit from the toolbar, click on the door icon, or press Ctl+x (the Control key and the x key together). If the program is in the middle of executing an analysis and you want to interrupt it, you can try pressing esc. This works only for iterative procedures such as MDS or TABU SEARCH. Otherwise press Control Alt Delete (simultaneously) and this brings up the operating system window click on the button marked task manager, you can now select UCINET and end the task. Some procedures offer a “Calculating …” dialogue box with a STOP uploads/Litterature/ users-guide.pdf