0

I have been working on a project that illustrates relationships between authors' articles and their respective citations (from other authors). Then, I created a matrix that shows edges between them.

Ultimately, we want to measure originality among all the articles, and we are open to additional suggestions on measuring originality.

Below is the code that I have already created (in RStudio using the bibtex package and igraph package):

data <- readFiles("network_science_450.bib") #read in data
convert<- convert2df(data, dbsource = "isi", format = "bibtex") #converted in a data frame

matrix <- cocMatrix(convert, Field = "CR", sep = ";")
sort(Matrix::colSums(matrix), decreasing = TRUE)[1:5]

NetMatrix <- biblioNetwork(convert, analysis = "coupling", network = "references", sep = ".  ")
NetMatrixTable <- as.matrix(NetMatrix, mode="directed", weighted=TRUE)

binary <- ifelse(NetMatrixTable>0,1,0) #converted into a binary matrix
as.matrix(binary)

We have created a binary matrix to represent all these relationships, but I was wondering if there is a better way to present our data. We have explored Hasse diagram as a possibility.

Our main problem is we cannot find a way to create an adjacency matrix to perform further analysis. We want to perform transitive reduction on the matrix.

Joseph Bosse
  • 1
  • 3
  • Without having data to use, it would seem that `cocMatrix` would generate an adjacency matrix (i.e. binary matrix representing which articles cite other articles). What downstream analyses. It would help if you could be more specific about the problem you are having and provide a sample of your data `dput(head(convert))` – emilliman5 Feb 21 '18 at 14:02

1 Answers1

0

I don't really understand your problem but it looks like you want to build a sociomatrix. If so, try: 

# citaton data
df <- data.frame(article = sample(LETTERS, 50, replace = TRUE),
                 cited_article = sample(LETTERS, 50, replace = TRUE))
## network creation
# 2-mode sociomatrix
df.2mode <- table(df)
df.2mode
# diag(df.2mode) <- 0

(A reproducible example is required for SO' questions)

nghauran
  • 6,648
  • 2
  • 20
  • 29