R Notebook

Creating the environment

library(tidyverse)
library(tosr)
library(bibliometrix)
library(lubridate)
library(igraph)
library(tidytext)
library(wordcloud)
library(rebus)
library(ggrepel) # improve donut visualization

Data getting

tree_of_science <- 
  tosr::tosR("ethno_education.bib",  # Create data from searches 
             "ethno_education.txt")

[1] 2

Converting your scopus collection into a bibliographic dataframe

Done!


Generating affiliation field tag AU_UN from C1:  Done!


Converting your wos collection into a bibliographic dataframe

Done!


Generating affiliation field tag AU_UN from C1:  Done!


 83 duplicated documents have been removed

Computing TOS SAP
Warning in TOS SAP
Computing TOS subfields

Table 1. Search Criteria

table_1 <- 
  tibble(wos = length(wos$SR), # Create a dataframe with the values.
         scopus = length(scopus$SR), 
         total = length(wos_scopus_tos$df$SR))
table_1

Figure 1. Languages

main_languages <- 
  wos_scopus_tos$df |> 
  select(LA) |> 
  separate_rows(LA, sep = "; ") |> 
  count(LA, sort = TRUE) |> 
  slice(1:5)

other_languages <- 
  wos_scopus_tos$df |> 
  separate_rows(LA, sep = "; ") |> 
  select(LA) |> 
  count(LA, sort = TRUE) |> 
  slice(6:n) |> 
  summarise(n = sum(n)) |> 
  mutate(LA = "OTHERS") |> 
  select(LA, n)

Warning in 6:n :
  numerical expression has 12 elements: only the first used

languages <- 
  main_languages |> 
  bind_rows(other_languages) |> 
  mutate(percentage = n / sum(n),
         percentage = round(percentage, 
                            digits = 2) ) |> 
  rename(language = LA) |>
  select(language, percentage, count = n)

languages

df <- languages |> 
  rename(value = percentage, group = language) |>
  mutate(value = value * 100) |> 
  select(value, group)

df2 <- df %>% 
  mutate(csum = rev(cumsum(rev(value))), 
         pos = value/2 + lead(csum, 1),
         pos = if_else(is.na(pos), value/2, pos))

ggplot(df, aes(x = 2 , y = value, fill = fct_inorder(group))) +
  geom_col(width = 1, color = 1) +
  coord_polar(theta = "y") +
  geom_label_repel(data = df2,
                   aes(y = pos, label = paste0(value, "%")),
                   size = 4.5, nudge_x = 1, show.legend = FALSE) +
   theme(panel.background = element_blank(),
        axis.line = element_blank(), 
        axis.text = element_blank(),
        axis.ticks = element_blank(),
        axis.title = element_blank(),
        plot.title = element_text(hjust = 0.5, size = 18)) +
  labs(title = "Languages") +
  guides(fill = guide_legend(title = "")) +
  theme_void() +
  xlim(0.5, 2.5)

Figure 2. Scientific Production

wos_anual_production <- 
  wos |> 
  select(PY) |> 
  count(PY, sort = TRUE) |> 
  na.omit() |> 
  filter(PY >= 2000,
         PY < year(today())) |> 
  mutate(ref_type = "wos")

scopus_anual_production  <- 
  scopus |> 
  select(PY) |> 
  count(PY, sort = TRUE) |> 
  na.omit() |> 
  filter(PY >= 2000,
         PY < year(today())) |>
  mutate(ref_type = "scopus")

total_anual_production <- 
  wos_scopus_tos$df |> 
  select(PY) |> 
  count(PY, sort = TRUE) |> 
  na.omit() |> 
  filter(PY >= 2000,
         PY < year(today())) |>
  mutate(ref_type = "total")

wos_scopus_total_annual_production <- 
  wos_anual_production |> 
  bind_rows(scopus_anual_production,
            total_anual_production) 

figure_2_data <- 
  wos_scopus_total_annual_production |> 
  mutate(PY = replace_na(PY, replace = 0)) |> 
  pivot_wider(names_from = ref_type, 
              values_from = n) |> 
  arrange(desc(PY))

figure_2_data 

wos_scopus_total_annual_production |> 
  ggplot(aes(x = PY, y = n, color = ref_type)) +
  geom_line() +
  labs(title = "Annual Scientific Production", 
       x = "years",
       y = "papers") +
  theme(plot.title = element_text(hjust = 0.5)) 

Table 2. Country production

data_biblio_wos <- biblioAnalysis(wos)

wos_country <- 
  data_biblio_wos$Countries |> 
  data.frame() |> 
  mutate(database = "wos") |> 
  select(country = Tab, papers = Freq, database ) |> 
  arrange(desc(papers)) 

data_biblio_scopus <- biblioAnalysis(scopus)

scopus_country <- 
  data_biblio_scopus$Countries |> 
  data.frame() |> 
  mutate(database = "scopus") |> 
  select(country = Tab, papers = Freq, database ) |> 
  arrange(desc(papers)) 

data_biblio_total <- biblioAnalysis(wos_scopus_tos$df)
  
total_country <- 
  data_biblio_total$Countries |> 
  data.frame() |> 
  mutate(database = "total") |> 
  select(country = Tab, papers = Freq, database ) |> 
  arrange(desc(papers)) 

wos_scopus_total_country <- 
  wos_country |> 
  bind_rows(scopus_country, 
            total_country) |> 
  mutate(country = as.character(country)) |> 
  pivot_wider(names_from = database, 
              values_from = papers) |> 
  arrange(desc(total)) |> 
  slice(1:10) |> 
  mutate(percentage = total / (table_1 |> pull(total)),
         percentage = round(percentage, digits = 2))

wos_scopus_total_country

Table 3. Author production

wos_authors <- 
  data_biblio_wos$Authors |> 
  data.frame() |> 
  rename(authors_wos = AU, papers_wos = Freq) |> 
  arrange(desc(papers_wos)) |> 
  slice(1:10) |> 
  mutate(database_wos = "wos")


scopus_authors <- 
  data_biblio_scopus$Authors |> 
  data.frame() |> 
  rename(authors_scopus = AU, papers_scopus = Freq) |> 
  arrange(desc(papers_scopus)) |> 
  slice(1:10) |> 
  mutate(database_scopus = "scopus")

total_authors <- 
  data_biblio_total$Authors |> 
  data.frame() |> 
  rename(authors_total = AU, 
         papers_total = Freq) |> 
  arrange(desc(papers_total)) |> 
  slice(1:10) |> 
  mutate(database_total = "total")

wos_scopus_authors <- 
  wos_authors |> 
  bind_cols(scopus_authors,
            total_authors)

wos_scopus_authors

Table 4. Journal production

wos_journal <- 
  wos |> 
  filter(str_detect(DT, "ARTICLE")) |> 
  select(journal = SO) |> 
  na.omit() |> 
  count(journal, sort = TRUE) |> 
  slice(1:20) |> 
  rename(publications = n) |> 
  mutate(database = "wos")

scopus_journal <- 
  scopus |> 
  filter(str_detect(DT, "ARTICLE")) |> 
  select(journal = SO) |> 
  na.omit() |> 
  count(journal, sort = TRUE) |> 
  slice(1:20) |> 
  rename(publications = n) |> 
  mutate(database = "scopus")

total_journal <- 
  wos_scopus_tos$df |> 
  filter(str_detect(DT, "ARTICLE")) |> 
  select(journal = SO) |> 
  na.omit() |> 
  count(journal, sort = TRUE) |> 
  slice(1:20) |> 
  rename(publications = n) |> 
  mutate(database = "total")

wos_scopus_total_journal <- 
  wos_journal |> 
  bind_rows(scopus_journal, 
            total_journal) |> 
  pivot_wider(names_from = database, 
              values_from = publications) |> 
  arrange(desc(total)) |> 
  slice(1:10) |> 
  mutate(percentage = total / table_1 |> pull(total),
         percentage = round(percentage, digits = 2))


wos_scopus_total_journal

Figure 3. Co-citation network

Author co-citation network

wos_scopus_author_metatag <- 
  metaTagExtraction(wos_scopus_tos$df, Field = "CR_AU")

wos_scopus_author_co_citation_matrix <- 
  biblioNetwork(M = wos_scopus_author_metatag, 
                analysis = "co-citation", 
                network = "authors")

plot_net_author_co_citation <- 
  networkPlot(wos_scopus_author_co_citation_matrix, 
              weighted=T, 
              n = 30, 
              Title = "Author Co-citation Network", 
              type = "fruchterman", 
              size=T,
              edgesize = 5,
              labelsize=0.7)

Author Collaboration network

wos_scopus_author_collab_matrix <- 
  biblioNetwork(M = wos_scopus_tos$df, 
                analysis = "collaboration", 
                network = "authors")

plot_author_collab <- 
  networkPlot(NetMatrix = wos_scopus_author_collab_matrix, 
              weighted=T, n = 30, 
              Title = "Author Collaboration Network", 
              type = "fruchterman", 
              size=T,
              edgesize = 5,
              labelsize=0.7)

Country Collaboration Network

wos_scopus_country_collab_matrix <- 
  biblioNetwork(M = wos_scopus_tos$df, 
                analysis = "collaboration", 
                network = "countries")

plot_country_collab <- 
  networkPlot(wos_scopus_country_collab_matrix, 
              weighted=T, n = 30, 
              Title = "Country Collaboration Network", 
              type = "fruchterman", 
              size=T,
              edgesize = 5,
              labelsize=0.7)

Keyword co-citation network

wos_scopus_keyword_co_occurrence_matrix <- 
  biblioNetwork(M = wos_scopus_tos$df, 
                analysis = "co-occurrences", 
                network = "keywords", 
                sep = ";")

plot_net_co_occurrence <- 
  networkPlot(wos_scopus_keyword_co_occurrence_matrix, 
              weighted=T, n = 30, 
              Title = "Keyword Co-occurrence Network", 
              type = "fruchterman", 
              size=T,
              edgesize = 5,
              labelsize=0.7)

Figure 4. Tree of Science

Tree of Science

tree_of_science

NULL

Clustering analysis

Finding the clusters

nodes <-  
  tibble(name = V(wos_scopus_tos$graph)$name) |> 
  left_join(wos_scopus_tos$nodes, 
            by = c("name" = "ID_TOS"))

wos_scopus_citation_network_1 <- 
  wos_scopus_tos$graph |> 
  igraph::set.vertex.attribute(name = "full_name", 
                               index = V(wos_scopus_tos$graph)$name, 
                               value = nodes$CITE)

nodes_1 <- 
  tibble(name = V(wos_scopus_citation_network_1)$name,
         cluster = V(wos_scopus_citation_network_1)$subfield,
         full_name = V(wos_scopus_citation_network_1)$full_name)
  
nodes_2 <- 
  nodes_1 |> 
  count(cluster, sort = TRUE) |> 
  mutate(cluster_1 = row_number()) |> 
  select(cluster, cluster_1)

nodes_3 <- 
  nodes_1 |> 
  left_join(nodes_2) |> 
  rename(subfield = cluster_1) |> 
  select(name, full_name, subfield)

Joining, by = "cluster"

edge_list <- 
  get.edgelist(wos_scopus_citation_network_1) |> 
  data.frame() |> 
  rename(Source = X1, Target = X2)

wos_scopus_citation_network <- 
  graph.data.frame(d = edge_list, 
                   directed = FALSE, 
                   vertices = nodes_3)

wos_scopus_citation_network |> 
  summary()

IGRAPH e6b120e UN-- 360 586 -- 
+ attr: name (v/c), full_name (v/c), subfield (v/n)

Choosing clusters

clusters <- 
  tibble(cluster = V(wos_scopus_citation_network)$subfield) |> 
  count(cluster, sort = TRUE)

clusters |> 
  ggplot(aes(x = reorder(cluster, n), y = n)) +
  geom_point() 

Removing not chosen clusters

wos_scopus_citation_network_clusters <- 
  wos_scopus_citation_network |> 
  delete.vertices(which(V(wos_scopus_citation_network)$subfield != 1 & # filter clusters 
                          V(wos_scopus_citation_network)$subfield != 2 &
                          V(wos_scopus_citation_network)$subfield != 3  &
                          V(wos_scopus_citation_network)$subfield != 4))

wos_scopus_citation_network_clusters |> 
  summary()

IGRAPH 29631ff UN-- 170 272 -- 
+ attr: name (v/c), full_name (v/c), subfield (v/n)

Cluster 1

nodes_full_data |> 
  filter(cluster == 1) |> 
  select(full_name) |> 
  mutate(full_name = str_extract(full_name, SPC %R%  # Regular expressions 
                                   one_or_more(WRD) %R% 
                                   SPC %R% 
                                   one_or_more(or(WRD, ANY_CHAR))),
         full_name = str_remove(full_name, OPEN_PAREN %R% 
                                  repeated(DGT, 4) %R% 
                                  CLOSE_PAREN %R%
                                  one_or_more(or(WRD,ANY_CHAR))),
         full_name = str_trim(full_name))  |> 
  unnest_tokens(output = word, input = full_name) |> # Tokenization
  anti_join(stop_words) |>  # Removing stop words
   filter(word != "doi",
         !str_detect(word, "[0-9]")) |>  # WoS data
  filter(word == str_remove(word, pattern = "citation"),
         word == str_remove(word, pattern = "research"),  # Words removed
         word == str_remove(word, pattern = "analysis"), 
         word == str_remove(word, pattern = "science"),
         word == str_remove(word, pattern = "scientometric"),
         word == str_remove(word, pattern = "education")) |>
  count(word, sort = TRUE) |> 
  with(wordcloud(word, 
                 n, 
                 random.order = FALSE, 
                 max.words = 50, 
                 colors=pal))

Joining, by = "word"

Cluster 2

nodes_full_data |> 
  filter(cluster == 2) |> 
  select(full_name) |> 
  mutate(full_name = str_extract(full_name, SPC %R%  # Regular expressions 
                                   one_or_more(WRD) %R% 
                                   SPC %R% 
                                   one_or_more(or(WRD, ANY_CHAR))),
         full_name = str_remove(full_name, OPEN_PAREN %R% 
                                  repeated(DGT, 4) %R% 
                                  CLOSE_PAREN %R%
                                  one_or_more(or(WRD,ANY_CHAR))),
         full_name = str_trim(full_name))  |> 
  unnest_tokens(output = word, input = full_name) |> 
  anti_join(stop_words) |>
  filter(word != "doi",
         !str_detect(word, "[0-9]")) |>  # WoS data
  filter(word == str_remove(word, pattern = "citation"),
         word == str_remove(word, pattern = "research"), 
         word == str_remove(word, pattern = "analysis"), 
         word == str_remove(word, pattern = "science"),
         word == str_remove(word, pattern = "scientometric"),
         word == str_remove(word, pattern = "education")) |>
  count(word, sort = TRUE) |> 
  with(wordcloud(word, 
                 n, 
                 random.order = FALSE, 
                 max.words = 50, 
                 colors=pal))

Joining, by = "word"

Cluster 3

nodes_full_data |> 
  filter(cluster == 3) |> 
  select(full_name) |> 
  mutate(full_name = str_extract(full_name, SPC %R%  # Regular expressions 
                                   one_or_more(WRD) %R% 
                                   SPC %R% 
                                   one_or_more(or(WRD, ANY_CHAR))),
         full_name = str_remove(full_name, OPEN_PAREN %R% 
                                  repeated(DGT, 4) %R% 
                                  CLOSE_PAREN %R%
                                  one_or_more(or(WRD,ANY_CHAR))),
         full_name = str_trim(full_name))  |> 
  unnest_tokens(output = word, input = full_name) |> 
  anti_join(stop_words) |>
   filter(word != "doi",
         !str_detect(word, "[0-9]")) |>  # WoS data 
  filter(word == str_remove(word, pattern = "citation"),
         word == str_remove(word, pattern = "research"), 
         word == str_remove(word, pattern = "analysis"), 
         word == str_remove(word, pattern = "science"),
         word == str_remove(word, pattern = "scientometric"),
         word == str_remove(word, pattern = "education")) |>
  count(word, sort = TRUE) |> 
  with(wordcloud(word, 
                 n, 
                 random.order = FALSE, 
                 max.words = 50, 
                 colors=pal))

Joining, by = "word"

Cluster 4

nodes_full_data |> 
  filter(cluster == 4) |> 
  select(full_name) |> 
  mutate(full_name = str_extract(full_name, SPC %R%  # Regular expressions 
                                   one_or_more(WRD) %R% 
                                   SPC %R% 
                                   one_or_more(or(WRD, ANY_CHAR))),
         full_name = str_remove(full_name, OPEN_PAREN %R% 
                                  repeated(DGT, 4) %R% 
                                  CLOSE_PAREN %R%
                                  one_or_more(or(WRD,ANY_CHAR))),
         full_name = str_trim(full_name))  |> 
  unnest_tokens(output = word, input = full_name) |> 
  anti_join(stop_words) |> 
   filter(word != "doi",
         !str_detect(word, "[0-9]")) |>  # WoS data
  filter(word == str_remove(word, pattern = "citation"),
         word == str_remove(word, pattern = "research"), 
         word == str_remove(word, pattern = "education"), 
         word == str_remove(word, pattern = "de"),
         word == str_remove(word, pattern = "la"),
         word == str_remove(word, pattern = "res")) |>
  count(word, sort = TRUE) |> 
  with(wordcloud(word, 
                 n, 
                 random.order = FALSE, 
                 max.words = 50, 
                 colors=pal))

Joining, by = "word"

Exporting files

write.csv(cluster_4_df, "cluster_4.csv") # Exporting cluster 4

Error in is.data.frame(x) : object 'cluster_4_df' not found