Exploration of co-citation networks

The objective of this document is to explore the co-citations among the FRB-MTE-OFB projects. The references were cleaned and completed in the scripts 01_clean_references.R and 02_fetch_openalex.R.

Estimate the distance among papers cited by projects using co-citations. The questions are: - how wide/narrow are the research projects? - how similar/interconnected are the projects? can we make group of projects based on their citations? - which are key work that are cited by many projects?

Nodes are all papers cited by the projects, and cited papers that are cited by at least two papers.

The Co-Citation Network maps the intellectual structure of a research field by analyzing patterns in how documents cite the same references. Two references are co-cited when they are both cited by the same document. The more frequently two references are co-cited, the stronger their intellectual relationship. This technique, introduced by Small (1973), is one of the foundational methods of bibliometric analysis.

Co-citation analysis rests on two complementary approaches:
- Co-citation (Small, 1973): Measures how often two references are cited together by subsequent publications. Highly co-cited references are perceived as intellectually related by the citing community. This method identifies the knowledge base of a field. - Bibliographic coupling (Kessler, 1963): Measures the overlap in the reference lists of two documents. Documents that share many cited references are likely working on similar topics. This method identifies current research fronts.

Both approaches build networks where: Nodes represent cited references (co-citation) or citing documents (bibliographic coupling) Edges represent the strength of co-citation or bibliographic coupling between two nodes Clusters represent distinct schools of thought, theoretical frameworks, or research traditions

General overview

devtools::load_all()
library(igraph) # for general network analysis
library(bibliometrix) # to build citation networks
library(visNetwork)
library(plotly)

out_data <- here::here("data", "derived-data")
# load reference list
ref <- read.csv(file.path(out_data, "mte_references_completed.csv"))
ref$Project <- firstup(ref$Project)
# remove references with no DOI
ref <- ref[!is.na(ref$DOI), ]

ref$Relation <- factor(
  ref$Relation,
  levels = c("Proposal", "Top5", "Output"),
  ordered = TRUE
)
ref <- ref[order(ref$Project, ref$Relation, ref$DOI, decreasing = TRUE), ]
# remove duplicates # sum(duplicated(ref))
ref <- ref[!duplicated(ref[, c("Project", "DOI")]), ]

meta <- readxl::read_xlsx(
  here::here("data", "raw-data", "mte", "Projets_FRB-MTECT-OFB.xlsx")
)
# all(meta$Acronyme %in% ref$Project)

# either based on previous script
# M <- readRDS(file.path(out_data, "mte_bibliometrix.rds"))
# dim(M) # 1035, 57
# or from shiny completed references
load(file.path(out_data, "mte_references_oa_shinycomplete.RData"))
# M$CR has short names that can't be linked back to id_oa
# how to get the short name
# dim(M) # 1047, 57
ref$Short <- M$SR[match(ref$DOI, M$DI)]
ref$FAut <- M$AU_CORR[match(ref$DOI, M$DI)]

Co-citation network

# get the co-citation network of references
NetMatrix <- bibliometrix::biblioNetwork(
  M,
  analysis = "coupling",
  network = "references",
  n = NULL,
  short = FALSE,
  sep = ";"
)
dim(NetMatrix) # 1035

[1] 1047 1047

# table(M$SR %in% dimnames(NetMatrix)[[1]])

# 46722 references with 627 original references

net1 <- igraph::graph_from_adjacency_matrix(
  NetMatrix,
  mode = "undirected",
  diag = FALSE,
  weighted = TRUE
)


# delete non connected literature
# net2 <- delete_vertices(net1, degree(net1) == 0)
net2 <- delete_vertices(net1, components(net1)$membership != 1)

# modularity
mod2 <- cluster_walktrap(net2)
clu2 <- membership(mod2)
sclu2 <- ifelse(clu2 <= (sum(table(clu2) > 1)), clu2, "isolated")
sclu2 <- as.factor(sclu2)
cclu <- colorspace::qualitative_hcl(n = nlevels(sclu2), palette = "Dark 3")
cclu[levels(sclu2) == "isolated"] <- "#808080" # grey
col2 <- cclu[sclu2]


deg2 <- degree(net2)

# V(net1)$size <- sqrt(degree(net1))
# V(netREF)$label <- ref$Short[match(V(net1)$name, ref$Short)]
# V(netREF)$year <- ref$Year[match(V(netREF)$name, ref$DOI)]
# V(netREF)$faut <- ref$FAut[match(V(netREF)$name, ref$DOI)]
# V(netREF)$fulltitle <- ref$Title[match(V(netREF)$name, ref$DOI)]
# V(netREF)$shorttitle <- substr(V(netREF)$fulltitle, 1, 25)
# V(netREF)$title <- paste(
#   V(netREF)$name,
#   V(netREF)$year,
#   V(netREF)$faut,
#   V(netREF)$shorttitle,
#   sep = "<br>"
# )
table(membership(mod2))

  1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19  20 
 23   2   3  15 131   5  65   8 161 145  12   8  30   5   2   3   4   4   3 132 
 21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40 
  4   3   2 108   3   3  15  20   3   3  17   5   4   2   2   2   1   1   1   1 
 41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 
  1   1   1   1   1   1   1   1   1   1   1   1   1   1   1   1   1   1   1   1 
 61  62  63  64  65  66  67  68  69 
  1   1   1   1   1   1   1   1   1 

V(net2)$group <- membership(mod2)
V(net2)$color <- col2

# check out other layout
visNetwork::visIgraph(
  net2,
  layout = "layout_with_kk",
  idToLabel = FALSE
) |>
  visOptions(
    highlightNearest = TRUE,
    selectedBy = list(variable = "group", multiple = T)
  )

The network has 990 nodes (one node = one article) and 6.5778\times 10^{4} edges (one edge = shared citation between two articles). The connectance is 0.134 and the modularity 0.28

centR <- data.frame(
  "betweenness" = betweenness(net2),
  "degree" = degree(net2),
  "pagerank" = page_rank(net2)$vector,
  "lab" = V(net2)$name
)

plot_ly(centR) |>
  add_markers(
    x = ~betweenness,
    y = ~pagerank,
    size = ~degree,
    text = ~lab,
    hoverinfo = "text"
  ) |>
  layout(title = "Reference centrality") |>
  config(
    modeBarButtons = list(list("toImage")),
    displaylogo = FALSE
  )

info <- apply(centR[, -4], 2, round, 5)
score <- apply(scale(sqrt(info)), 1, mean)
info <- as.data.frame(info)
DT::datatable(info[order(score, decreasing = TRUE)[1:50], ])

Per project

info <- c()
for (i in unique(ref$Project)) {
  subi <- delete_vertices(
    net2,
    !V(net2)$name %in% ref$Short[ref$Project == i]
  )
  ni <- c("P" = i, "N" = vcount(subi), "C" = edge_density(subi))
  info <- rbind(info, ni)
}
info <- as.data.frame(info)
info$N <- as.numeric(info$N)
info$C <- as.numeric(info$C)
plot_ly(info) |>
  add_markers(
    x = ~N,
    y = ~C,
    text = ~P,
    hoverinfo = "text"
  ) |>
  layout(title = "Project bibliographic coupling") |>
  config(
    modeBarButtons = list(list("toImage")),
    displaylogo = FALSE
  )