Future plots/visualizations

[TNRiley]

This board is dedicated to brainstorming future plots/visualizations. If you've got an idea, feel free to post. If you've got an example of a visualization you think would be useful, even better!

[TNRiley]

One visualization the team talked about, but have not had time to explore is a sankey or an alluvial plot. They are similar in structure and there are a few packages that I'm aware of that produce these types of plots (ggsankey, ggalluvial, networkD3).

Currently only cite_source and cite_label are used in our plots. Outside of a wildly complex venn diagram, this plot is one of the only ways (that I have been able to think of) that we could incorporate all three variables cite_source, string, and label.

One way the three tags could be integrated using these plots..... Search strings feed into resources. Those then feed into either included or excluded at ti/ab, and finally included/excluded at full text.

Here is a very basic example of how this might look.

If you want to play around with this, here is the file you can use to upload at sankeymatic.com
sankeymatic_20230312_165214_source.txt

[DrMattG]

R code to create the same sankey

suppressPackageStartupMessages(library(networkD3,warn.conflicts = FALSE))
#> Warning: package 'networkD3' was built under R version 4.1.3
suppressPackageStartupMessages(library(tidyverse,warn.conflicts = FALSE))
#> Warning: package 'tidyverse' was built under R version 4.1.3
#> Warning: package 'ggplot2' was built under R version 4.1.3
#> Warning: package 'tibble' was built under R version 4.1.3
#> Warning: package 'tidyr' was built under R version 4.1.3
#> Warning: package 'readr' was built under R version 4.1.3
#> Warning: package 'purrr' was built under R version 4.1.3
#> Warning: package 'dplyr' was built under R version 4.1.3
#> Warning: package 'stringr' was built under R version 4.1.3
#> Warning: package 'forcats' was built under R version 4.1.3
#> Warning: package 'lubridate' was built under R version 4.1.3

nodes<-data.frame(name=c("WoS1", "WoS2", "WoS", 
                         "ASFA1", "ASFA2","ASFA",
                         "Scopus1", "Scopus2", "Scopus",
                         "Included_TI_AB", 
                         "Excluded_TI_AB", "Final_Included",
                         "Final_Excluded"), id=seq(0,12,1))

links<-data.frame(source=c(0, 1, 3,
                           4, 6, 7,
                           2, 2,
                           5,5,
                           8, 8,
                           9,
                           9), target=c(
                             2, 2,5,5,
                             8, 8, 9,
                             10, 9,
                             10,9,
                             10, 11, 12
                           ), value=c(100,150,100,150,100,150,
                                      50,200,50,200,50,200,50,100))


p <- sankeyNetwork(Links = links, Nodes = nodes, Source = "source",
                   Target = "target", Value = "value", NodeID = "name",
                   units = "papers", fontSize = 12, nodeWidth = 30)
p

^{Created on 2023-03-15 with reprex v2.0.2}

[TNRiley]

Thanks for this @DrMattG - I've put this together, but running into an issue. I want Method 1 and Method 2 to be in the second column. Not sure if this is possible without something feeding into them?

#furhter work needed to integrate Unique_citations into this plot, as this is very manual, but great as an example as to how the three custom fields can be visualized. In this case Strings/Sources/Timeline (Inclusion/Exclusion)

# Define nodes
nodes <- data.frame(
  name = c("AG 1", "AG 2", "AGRIS", 
           "CAB 1", "CAB 2", "CAB",
           "EL 1", "EL 2", "EconLit", 
           "WoS 1", "WoS 2", "WoS", 
           "Method 1", "Method 2", 
           "TI/AB Included", "Final Included", "Excluded"),
  id = 0:16
)

# Define links
links <- data.frame(
  source = c(0, 1, 3, 4, 6, 7, 9, 10),
  target = c(2, 2, 5, 5, 8, 8, 11, 11),
  value = c(100, 150, 100, 150, 100, 150, 100, 100)
)

# Add links for the second column nodes (including Method 1 and Method 2) flowing into TI/AB Included and Excluded
links <- rbind(links, data.frame(
  source = c(2, 5, 8, 11, 2, 5, 8, 11, 12, 12, 13, 13),
  target = c(14, 14, 14, 14, 16, 16, 16, 16, 14, 16, 14, 16),
  value = c(50, 50, 50, 50, 200, 200, 200, 150, 50, 50, 50, 50)
))

# Add links for TI/AB Included flowing into Final Included and Excluded
links <- rbind(links, data.frame(
  source = c(14),
  target = c(15),
  value = c(100)
))

links <- rbind(links, data.frame(
  source = c(14),
  target = c(16),
  value = c(200)
))


# Create Sankey diagram
sankey<-sankeyNetwork(
  Links = links,
  Nodes = nodes,
  Source = "source",
  Target = "target",
  Value = "value",
  NodeID = "name",
  units = "",
  fontSize = 12,
  nodeWidth = 30,
   iterations = 0
)
sankey

[TNRiley]

Another one - this is a table that should be pretty useful in seeing the journal titles with the highest number of unique citations per source. Consider adding these functions to CS to make this easy. Table example and code below. This example is based on three sources DIM, LENS, WoS. Idea based on @rootsandberries initial table on comparing database topic coverage.

# This function takes a data frame as input, and returns a processed data frame
process_journals <- function(data) {
  data %>%
    group_by(journal) %>%          # Group the data by journal
    summarise(count = n()) %>%
    dplyr::filter(journal != "") %>%  # Filter out any blank journal titles
    arrange(desc(count))             # Arrange the journals in descending order of count
}


# This function takes a data frame and a caption as input, and returns a gt table
display_gt_table <- function(data, caption) {
  data %>%
    gt() %>%
    tab_header(title = caption) %>%  # Set the table title to the provided caption
    cols_label(journal = "Journal", count = "Unique Citations") %>%  # Set column labels
    tab_options(
      table.width = px(600),
      table.font.size = px(12),
      heading.title.font.size = px(16),
      heading.subtitle.font.size = px(14)
    )                                   # Set table styling options
}

# Process and display top 10 journals for each source
DIM_journals <- process_journals(unique_DIM)  # Process DIM journals
display_gt_table(DIM_journals[1:10, ], "Top 10 Journals by Unique Citations DIM")  # Display gt table for DIM_journals

LENS_journals <- process_journals(unique_LENS)  # Process LENS journals
display_gt_table(LENS_journals[1:10, ], "Top 10 Journals by Unique Citations LENS")  # Display gt table for LENS_journals

WoS_journals <- process_journals(unique_WoS)  # Process WoS journals
display_gt_table(WoS_journals[1:10, ], "Top 10 Journals by Unique Citations WoS")  # Display gt table for WoS_journals

[TNRiley]

I've developed a new table that I've added to the plot sandbox vignette. This goal of this table is to show users the number of initial records that were imported from each source and the number of unique records after internal deduplication. I've also included the number of unique records that source contributed after deduplication across all sources, but that might no be necessary as that number will be reported on the larger citation summary table.

I'd love to review this with everyone and then get a little hand holding as I add the count_databases function to the package.

[TNRiley]

complete and integrated as record_counts_table

[TNRiley]

New idea -> Create a horizontal bar chart showing each source.
The bars would be stacked, and broken down by
unique, overlap w/ 1 source, overlap w/2 sources, overlap w/ 3+ sources

[TNRiley]

another potential option would be one bar for unique records broken down by source and then each source would have a bar with that shows the breakdown of their crossover records and how many sources they were in 1, 2, 3, etc.