Table of Contents

Analysis: authors

S2ORC metadata networks

WA indegrees

From the list of 100 largest entries in the indegrees vector

WA indegrees

we see:

Solutions: 

      firstAu = Au[0].strip() if len(Au)>0 else "Anonymous" 
      name = firstAu.split(",")[0] if len(firstAu)>0 else "Anonymous"

Distribution of the number of authors

In Pajek 

Select network WA
Network/2-mode Network/Partition 2-mode
Network/Create Vector/Centrality/Degree/Output
Operations/Vector+Partition/Extract Subvector [1]
Save vector [Aout.vec]

and in R 

> wdir <- "C:/Users/batagelj/Documents/2020/corona/MetaTit"
> setwd(wdir)
> na <- read.table('Aout.vec',skip=1)[,1]
> length(na)
[1] 375094
> f <- table(na)
> length(f)
[1] 163
> n <- as.integer(names(f))
> plot(n,f)
> plot(n,f,log='xy',pch=16)

Normalized collaboration network

WAn = norm(WA). Cn = WAnT * WAn (see BC).

In Pajek (norm2.mcr from Biblio) 

Select network WA
Macro/Play [norm2.mcr] [375094]
Network/2-mode Network/Transpose 2-mode
Select normalized network as Second network
Networks/Multiply Networks
Network/Create Vector/Get Loops

Top 100 authors for fractional contribution

Cn[a,a] = fractional contribution of the author a to coauthored works

Top 100 authors for fractional contribution

We see:

Ps cores

First transform Cn to undirected network. 

Select network Cn
Network/Create New Network/Transform/Remove Loops
Network/Create New Network/Transform/Arcs to Edges/Bidirected Only/Sum ValuesNetwork/Create Vector/Generalized Cores/Sum/All
Info Ps vector [+200]
Vector/Make Partition/by Intervals/Selected [5.0]
Operations/Network+Partition/Extract/Subnetwork Induced ... [2]

Each node of the Ps core has the internal sum of weights (weighted degree) at least 5: Ps core at level 5.0

We see:

Simple islands

Single peak islands - single topic. 

Network/Create Partition/Islands/Line Weights (simple) [2,100] [yes]
Operations/Network+Partition/Extract/Subnetwork Induced ... [92040-92158]
Partition/Canonical Partition/with Decreasing Freq
Info Partition [+30]
Operations/Network+Partition/Extract/Subnetwork Induced ... [1-18]

There are 92158 simple islands with their size in [2,100]

Top Cn simple islands for heights

Top Cn simple islands for size

We see:

Solutions:

Islands

Network/Create Partition/Islands/Line Weights [2,50] [yes]
Save Islands partition to the file NcIsla2.clu
Save Islands heights vector to the file NcIslaH2.vec

There are 81934 islands with size in [2,50]. For each island we know its size and its height. I selected 50 top islands for the weight = sqrt(size-1)*height. 

wdir <- "C:/Users/batagelj/Documents/2020/corona/MetaTit"
setwd(wdir)
h <- read.table('NcIslaH2.vec',skip=1)[,1]
C <- read.table('NcIsla2.clu',skip=1)[,1]
N <- length(C); k <- max(C); H <- rep(-1,k)
for(i in 1:N){ j <- C[i]
   if(j > 0){if(H[j] < 0) H[j] <- h[i]}
}
w <- sqrt(s[2:(k+1)]-1)*H
r <- cbind(s[2:(k+1)],H,w)
q <- order(w,decreasing=TRUE)
R <- r[q,]
> R[1:50,]
                 H        w
81934  2 7.2500000 7.250000
77577 42 0.4952564 3.171188
81933  2 3.0000000 3.000000
...
76851 42 0.2630385 1.684269
62911 49 0.2405513 1.666588
62883 50 0.2361111 1.652778
71695 44 0.2500000 1.639360
...
81913  2 1.5000000 1.500000
81914  2 1.5000000 1.500000
> paste(row.names(R[1:50,]),collapse=",")
[1] "81934,77577,81933,81932,81931,81924,81918,81927,81928,81929,81930,77063,76992,81903,79495,81925,81926,
     81901,76826,76782,67555,81916,81917,81919,81920,81921,81922,81923,76851,62911,62883,71695,62902,62930,
     62834,62835,62316,64394,65412,81904,81905,81906,81907,81908,81909,81910,81911,81912,81913,81914"
Operations/Network+Partition/Extract/Subnetwork Induced ... [list from R]
Network/Create New Network/Transform/Line Values/Absolute+Sqrt

Top Cn islands

For visualization because of the large range, the weights on edges were transformed using sqrt.

The picture has to be improved manually.

We see: