Presentation on theme: "Pajek. Pajek is a program, for Windows, for analysis and visualization of large networks having some thousands or even millions of vertices. In Slovenian."— Presentation transcript:
Pajek is a program, for Windows, for analysis and visualization of large networks having some thousands or even millions of vertices. In Slovenian language the word pajek means spider.
Pajek should provide tools for analysis and visualization of such networks: collaboration networks, organic molecule in chemistry, protein-receptor interaction networks, genealogies, Internet networks, citation networks, diffusion (AIDS, news, innovations) networks, data-mining (2-mode networks), etc. See also collection of large networks at:
to support abstraction by (recursive) decomposition of a large network into several smaller networks that can be treated further using more sophisticated methods; to provide the user with some powerful visualization tools; to implement a selection of efﬁcient (subquadratic) algorithms for analysis of large networks.
network – main object (vertices and lines - arcs, edges): graph, valued network, 2-mode or temporal network partition Nominal property of vertices. Default extension:.clu vector numerical property of vertices. Default extension:.vec permutation reordering of vertices. Default extension:.per cluster subset of vertices (e.g. a class from partition). Default extension:.cls. hierarchy hierarchically ordered clusters and vertices. Default extension:.hie
Network can be defined in different ways on input file. Look at three of them: 1. List of neighbours (Arcslist / Edgeslist)(see test 1.net)see test 1.net *Vertices 5 1 ”a” 2 ”b” 3 ”c” 4 ”d” 5 ”e” *Arcslist *Edgeslist 1 5
Data must be prepared in an input (ASCII) file. Program NotePad can be used for editing. Much better is a shareware editor, TextPad.TextPad Words, starting with *, must always be written in first column of the line. They indicate the start of a definition of vertices or lines. Using *Vertices 5 we define a network with 5 vertices. This must always be the first statement in definition of a network. Definition of vertices follows after that – to each vertex we give a label, which is displayed between “ and ”. Using *Arcslist, a list of directed lines from selected vertices are declared (1 2 4 means, that there exist two lines from vertex 1, one to vertex 2 and another to vertex 4). Similarly *Edgeslist, declares list of undirected lines from selected vertex. In the file no empty lines are allowed – empty line means end of network.
2. Pairs of lines (Arcs / Edges) (see test 2.net)see test 2.net *Vertices 5 1 ”a” 2 ”b” 3 ”c” 4 ”d” 5 ”e” *Arcs *Edges 1 5 1
Directed lines are defined using *Arcs, undirected lines are defined using *Edges. The third number in rows defining arcs/edges gives the value/weight of the arc/edge. In the previous format (Arcslist / Edgeslist) values of lines are not defined the format is suitable only if all values of lines are 1. If values of lines are not important the third number can be omitted (all lines get value 1). In the file no empty lines are allowed – empty line means end of network.
3.Matrix (see test 3.net)see test 3.net *Vertices 5 1 ”a” 2 ”b” 3 ”c” 4 ”d” 5 ”e” *Matrix
In this format directed lines (arcs) are given in the matrix form (*Matrix). If we want to transform bidirected arcs to edges we can use “Network>create new network>Transform>Arcs to Edges>Bidirected only”
Additionally, Pajek enables precise definition of elements used for drawing networks (coordinates of vertices, shapes and colors of vertices and lines,...). Example: (see test 4.net)see test 4.net *Vertices 5 1 “a” box 2 “b” ellipse 3 “c” diamond 4 “d” triangle 5 “e” empty...
Layout of networks Energy: The network is presented like a physical system, and we are searching for the state with minimal energy Kamada-Kawai: using separate components, you can tile connected components in a plane Fruchterman-Reingold: draw in a plane or space and selecting the repulsion factor Eigen Values: Selecting 2 or 3 eigenvectors to become the coordinates of vertices. Can obtain nice pictures
Partitions are used to describe nominal properties of vertices. e.g., 1-men, 2-women Definition in input file (see test.clu)see test.clu *Vertices
Vectors are used to describe numerical properties of vertices (e.g., centralities). Definition in input file (see test.vec) *Vertices
It is time consuming to load objects one by one. Therefore it is convenient to store all data in one file, called Pajek project file (.paj). (see test.paj)see test.paj Project files can be produced manually by using “File>Pajek Project File>Save” To load objects stored in Pajek project file select “File>Pajek Project File>Read”
Commands are put to menu according to the following criterion: commands that need only a network as input are available in menu Net, commands that need as input two networks are available in menu Networks, commands that need as input two objects (e. g., network and partition) are available in menu Operations, commands that need only a partition as input are available in menu Partition...
Local view is obtained by extracting sub-network induced by selected cluster of vertices. Global view is obtained by shrinking vertices in the same cluster to new (compound) vertex. In this way relations among clusters of vertices are shown. Combination of local and global view is contextual view: Relations among clusters of vertices and selected vertices are shown.
Import and export in 1994 among 80 countries are given. They is given in 1000$. (See Country_Imports.net)See Country_Imports.net Partition according to continents (see Country_Continent.clu)see Country_Continent.clu 1 – Africa, 2 – Asia, 3 – Europe, 4 – N. America, 5 – Oceania, 6 – S. America. Operations>Extract from Network>Partition Operations>Shrink Network>Partition
Operations>Extract from Network>Partition
Network>Create New Network>Transform>Remove>Lines with value>lower than (340000)
Download The latest version of Pajek is freely available, for non-commercial use, at its home page: Text file into Pajek WoS to Pajek Tutorial Exploratory Social Network Analysis with Pajek visit Pajek wiki for more information
The download link: The new tutorial slides: s:wos2pajek07.pdf s:wos2pajek07.pdf
Download from: Unpack it and copy ‘montylingua-2.1’ to C:\Python26\Lib\site-packages Set up a new environment variable named ‘MONTYLINGUA’ and set the variable value as c:\Python26\Lib\site-packages\MontyLingua- 2.1\Python
Download the latest version of WoS2Pajek. Unpack it, and double click on WoS2Pajek.py to show the main interface of program:
You can also put all wos files in a folder
The current version of WoS2Pajek requires 7 parameters to be given by the user: MontyLingua directory: path to the directory in which the MontyLingua package is installed; project directory: where the output files are saved; WoS file; maxnum – estimate of the number of all vertices (number of records+number of cited Works) –30*number of records; step – prints info about each k*step record as a trace; step= 0– no trace. use ISI name / short name; make a clean WoS file without duplicates; boolean list[DE, ID, TI, AB] specifying which fields are sources of keywords.
Network/Info/General Network/Create New Network/Transform/Remove/Loops Network/Create New Network/Transform/Remove/Multiple lines/Single line
Paper citation network Questions What are highly cited articles? The diameter of the network? What are the major clusters? More questions?
The network is transformed into an ordinary network, where the vertices are elements from the first subset, using “Network>2 mode network>2-Mode to 1-Mode>Rows”.
If we want to get a network with elements from the second subset we use “Network>2 mode network>2-Mode to 1-Mode>Columns”.
Basic information can be obtained by “Network>Info>General” which is available in the main window of the program. We get number of vertices number of arcs, number of directed loops number of edges, number of undirected loops density of lines Additionally we must answer the question: Input 1 or 2 numbers: +/highest, -/lowest where we enter the number of lines with the highest/lowest value or interval of values that we want to output. If we enter 10, 10 lines with the highest value will be displayed. If we enter -10, 10 lines with the lowest value will be displayed. If we enter 3 10, lines with the highest values from rank 3 to 10 will be displayed.
Load metformin network to Pajek
Ding, Y., Song, M., Han, J., Yu, Q., Yan, E., Lin, L., & Chambers, T. (2013). Entitymetrics: Measuring the impact of entities. PLoS One, 8(8): Entitymetrics is defined as using entities (i.e., evaluative entities or knowledge entities) in the measurement of impact, knowledge usage, and knowledge transfer, to facilitate knowledge discovery.
Network/Create New Network/SubNetwork with Paths/Info on Diameter Pajek returns only the two vertices that are the furthest away.
Strongly connected components Network>Create Partition>Components>Strong Weakly connected components Network>Create Partition>Components>Weak Result is represented by a partition vertices that belong to the same component have the same number in the partition. Example component.net
Go to partition weak component, Partition>make network>random network>Input Visualize the new random network
A cut-vertex is a vertex whose deletion increases the number of components in the network. A bi-component is a component of minimum size 3 that does not contain a cut-vertex.
Network/Create New Network/......with Bi-Connected Components stored as Relation Numbers Bicommponents are stored in hierarchy Load USAir97.net Get bicomponents with (14 of them) with component size >3
The largest component is 244 airports
Hierarchy>Extract Cluster (13), then result is stored in cluster Draw the cluster
A subset of vertices is called a k-core if every vertex from the subset is connected to at least k vertices from the same subset. K-Cores can be computed using “Network>Create Partitions>K-Core” and selecting Input, Output or All core. Result is a partition: for every vertex its core number is given. In most cases we are interested in the highest core(s) only. The corresponding subnetwork can be extracted using “Operations>Extract from Network>Partition” and typing the lower and upper limit for the core number. Example See k_core.net
How three nodes are connected Calculation of local Clustering Coefﬁcients: Network>Create Vector>Clustering Coefﬁcients>CC1 K_core.net
How nodes are connecting different clusters Betweenness centrality Network>Create vector>Centrality>Betweenness
The betweenness centrality value for each node
Closeness centrality Network>Create Vector>Centrality>Closeness Showing how one node is close to all other nodes in the network
Network/Create New Network/SubNetwork with Paths/.....One Shortest Path between Two Vertices Enter two vertices Forget values on lines Yes, if searching for the shortest path is based on lengths No, if searching for the shortest path is based o vlaue of lines Identify vertices in source network No Result will be a new subnetwork containing the two selected vertices Layout>Energy>Kamada Kawai>Fix first and last
Network/Create New Network/SubNetwork with Paths/.....One Shortest Path between Two Vertices ( )