+
+
diff --git a/content/contributor-analysis/network-graph.Rmd b/content/contributor-analysis/network-graph.Rmd
index 1ab83d430d8..1aabee399a9 100644
--- a/content/contributor-analysis/network-graph.Rmd
+++ b/content/contributor-analysis/network-graph.Rmd
@@ -6,9 +6,26 @@ output:
keep_md: FALSE
---
-The interactive network visualization below shows the connections between FORRT contributors based on their collaborative work across different projects. Each node represents a contributor, and the connections (edges) represent shared project participation.
+The interactive network visualization below shows the connections between FORRT contributors based on their collaborative work across different projects.
+
+**How to read the graph:**
+
+- **Nodes** represent contributor names. Each node is connected to other people involved in the same project(s).
+- **Edges** (lines) represent connections between contributors who worked on the same project.
+
+Use the dropdown menus below to search for a specific contributor by name or to filter the network by project. Hovering over a node highlights its connections and shows which projects that person contributed to. Clicking on a node highlights it in red, and clicking on an edge highlights the connection between two contributors. You can also zoom in and out using the scroll wheel or the navigation buttons.
+
+
```{r setup, include=FALSE}
@@ -40,168 +58,133 @@ library(tidyr)
library(igraph)
library(googlesheets4)
library(stringr)
+library(here)
+library(readr)
```
```{r data-processing}
-# Google Sheet URLs
-google_sheet_url_automation <- "https://docs.google.com/spreadsheets/d/1MUD54FQUhfcBKrvr5gCYoh2wgbJ6Lf7oAJRAqsQ-Nag/edit"
-google_sheet_url_leads <- "https://docs.google.com/spreadsheets/d/1roy-sZTxyXENA5c5IIV7IIemYvzbzs7zojUN2yIpi58/edit?gid=0#gid=0"
-# Get sheet names
-gs4_deauth()
-sheets <- sheet_names(google_sheet_url_automation)
-
-# Exclude the first four Automation Source sheets
-relevant_sheets <- sheets[-c(1,2,3,4)]
-
-# Exclude 'full' sheets
-relevant_sheets <- relevant_sheets[!grepl(" - full$", relevant_sheets)]
-
-# Read all relevant sheets into a list and remove rows without last names
-internal_links <- read_sheet(google_sheet_url_automation, sheet = "INTERNAL LINKS") %>%
- select(`Project Name`) # Extract project names
-
-data_list <- lapply(relevant_sheets, function(sheet) {
- # Read the content of the current sheet
- df <- read_sheet(google_sheet_url_automation, sheet = sheet)
- # Remove rows where the `Surname` is missing or empty
- df <- df %>% filter(!is.na(`Surname`) & `Surname` != "")
- # Match the `Project Name` using the sheet name (assuming sheet name is the identifier)
- project_name <- internal_links$`Project Name`[internal_links$`Project Name` == sheet]
- # If a match is found, assign the `Project Name`, else set to NA
- if (length(project_name) > 0) {
- df <- df %>% mutate(`Project Name` = project_name)
- } else {
- df <- df %>% mutate(`Project Name` = NA)
- }
- df
-})
-
-# Combine all sheets into one dataframe
-dt <- bind_rows(data_list, .id = "Source_Sheet")
-
-# Move project name column to first position
-dt <- dt %>%
- select(`Project Name`, everything())
-
-# Remove PM column as it is unnecessary and might cause problems with binding
-dt <- dt %>%
- select(-`Project Managers`)
+# Read role column mappings (same source as tenzing.py)
+fields_url <- "https://docs.google.com/spreadsheets/d/e/2PACX-1vT_IaXiYtB3iAmtDZ_XiQKrToRkxOlkXNAeNU2SIT_J9PxvsQyptga6Gg9c8mSvDZpwY6d8skswIQYh/pub?output=csv&gid=277271370"
+role_cols <- read_csv(fields_url, show_col_types = FALSE) |> pull(Fields)
+
+# Read contributor data from cache (generated by tenzing.py)
+cache_path <- here("scripts", "forrt_contribs", "contributors_cache.csv")
+credit_roles <- read_csv(cache_path, show_col_types = FALSE)
+
```
```{r leads-data}
# Read Leads Tenzing data
-leads_sheets <- sheet_names(google_sheet_url_leads)
-relevant_leads_sheets <- leads_sheets[-c(2)]
-leads_df <- read_sheet(google_sheet_url_leads)
+google_sheet_url_leads <- "https://docs.google.com/spreadsheets/d/1roy-sZTxyXENA5c5IIV7IIemYvzbzs7zojUN2yIpi58/edit?gid=0#gid=0"
+gs4_deauth()
-# Rename columns and select relevant ones
-leads_df <- leads_df %>%
+lead_roles <- read_sheet(google_sheet_url_leads) |>
rename(
`Project Name` = `FORRT project(s)`,
`ORCID iD` = `ORCID`
- ) %>%
- select(`First name`, `Middle name`, `Surname`, `Project Name`, `Role`, `ORCID iD`)
+ ) |>
+ select(`First name`, `Middle name`, `Surname`, `Project Name`, `Role`, `ORCID iD`) |>
+ mutate(has_role = TRUE) |>
+ distinct(`First name`, `Middle name`, `Surname`, `Project Name`, `ORCID iD`, `Role`, .keep_all = TRUE) |>
+ pivot_wider(names_from = `Role`, values_from = `has_role`, values_fill = list(has_role = FALSE)) |>
+ # Remove FORRT-wide organizational roles (not project-level contributions)
+ select(-Director, -`Operations Coordinator`)
-# Create a column to mark the presence of a role and group by individual
-leads_df <- leads_df %>%
- mutate(has_role = TRUE) %>%
- distinct(`First name`, `Middle name`, `Surname`, `Project Name`, `ORCID iD`, `Role`, .keep_all = TRUE)
+# Get leadership role columns (everything except ID columns)
+id_cols <- c("First name", "Middle name", "Surname", "Project Name", "ORCID iD")
+leads_role_cols <- setdiff(names(lead_roles), id_cols)
-# Pivot the data to create a column for each leadership role type, removing director
-leads_df <- leads_df %>%
- pivot_wider(names_from = `Role`, values_from = `has_role`, values_fill = list(has_role = FALSE)) %>%
- select(-Director, -`Operations Coordinator`)
+# TODO: Including leadership roles (all_role_cols) in the network pivot changes the
+# graph topology significantly, which breaks the visIgraphLayout centering and makes
+# the network unreadable. For now, only CRediT roles (role_cols) are used for the
+# network. A future fix should integrate leadership roles without breaking the layout.
+# all_role_cols <- union(role_cols, leads_role_cols)
-# Combine leads tenzing rows with automation source
-dt <- bind_rows(dt, leads_df)
+# Combine leads tenzing rows with automation source (CRediT roles only)
+credit_roles <- bind_rows(credit_roles, lead_roles)
```
```{r data-cleaning}
-# Trim names and ORCIDs to ensure inconsistent Tenzing entries are not counted separately
-dt <- dt %>%
+# Trim names and ORCIDs to ensure inconsistent Tenzing entries are not counted separately
+credit_roles <- credit_roles |>
mutate(
`First name` = str_trim(str_replace_all(`First name`, "\\*", "")),
- `Middle name` = str_trim(str_replace_all(`Middle name`, "\\*", "")) %>% str_sub(1, 1),
- `Surname` = str_trim(str_replace_all(`Surname`, "\\*", "")),
+ `Middle name` = str_trim(str_replace_all(`Middle name`, "\\*", "")) |> str_sub(1, 1),
+ Surname = str_trim(str_replace_all(Surname, "\\*", "")),
`ORCID iD` = str_trim(str_remove(`ORCID iD`, "https://orcid.org/"))
)
-
-# Ensure Conceptualization is logical
-dt$Conceptualization <- as.logical(dt$Conceptualization)
```
```{r network-data-preparation}
# Process data for network
-dt_long <- dt %>%
- pivot_longer(cols = c("Conceptualization", "Data curation", "Formal analysis", "Funding acquisition", "Investigation", "Methodology", "Project administration", "Resources", "Software", "Supervision", "Validation", "Visualization", "Writing - original draft", "Writing - review & editing"),
- names_to = "Role",
- values_to = "has_role") %>%
- filter(has_role == TRUE) %>%
- rowwise() %>%
- mutate(Contributor = paste(na.omit(c(`Surname`, `First name`)), collapse = " ")) %>%
- mutate(Lead = if_else(Role %in% c("lead", "co-lead"), Role, "other"))
-
-# Get co-occurences
-Contributor_Project <- xtabs(~ `Project Name` + Contributor,
- data=dt_long %>%
- select(`Project Name`, Contributor) %>%
- unique(),
- sparse = TRUE)
-
-Contributor_occur <- crossprod(Contributor_Project, Contributor_Project)
+credit_roles_long <- credit_roles |>
+ pivot_longer(
+ cols = any_of(role_cols),
+ names_to = "Role",
+ values_to = "has_role"
+ ) |>
+ filter(has_role == TRUE) |>
+ unite(Contributor, Surname, `First name`, sep = " ", remove = FALSE, na.rm = TRUE)
+ # TODO: uncomment to style lead/co-lead nodes differently
+ # mutate(Lead = if_else(Role %in% c("lead", "co-lead"), Role, "other"))
+
+# Get co-occurrences
+contributor_project_matrix <- credit_roles_long |>
+ distinct(`Project Name`, Contributor) |>
+ xtabs(~ `Project Name` + Contributor, data = _, sparse = TRUE)
+
+cooccurrence_matrix <- crossprod(contributor_project_matrix, contributor_project_matrix)
```
```{r create-nodes}
# Create nodes
-Contributor_nodes_n <- dt_long %>%
- group_by(Contributor) %>%
- count(Contributor)
-
-colnames(Contributor_nodes_n) <- c("id", "value")
-
-ContributorGroups_nodes <- dt_long %>%
- select(Contributor, `Project Name`) %>%
- unique() %>%
- group_by(Contributor) %>%
- mutate(Projects = paste0(`Project Name`, collapse = ",")) %>%
- select(Contributor, Projects) %>%
- unique()
+node_counts <- credit_roles_long |>
+ count(Contributor) |>
+ rename(id = Contributor, value = n)
-colnames(ContributorGroups_nodes) <- c("id", "projects")
+node_projects <- credit_roles_long |>
+ distinct(Contributor, `Project Name`) |>
+ summarise(projects = str_c(`Project Name`, collapse = ","), .by = Contributor) |>
+ rename(id = Contributor)
-Contributor_nodes <- left_join(Contributor_nodes_n, ContributorGroups_nodes)
-Contributor_nodes$title <- paste0(Contributor_nodes$id,"
+
diff --git a/layouts/partials/network-graph.html b/layouts/partials/network-graph.html
index b98b466f2a6..a18ea2a83c5 100644
--- a/layouts/partials/network-graph.html
+++ b/layouts/partials/network-graph.html
@@ -5661,6 +5661,9 @@
{{ .Content }}
-
+
-{{ partial "network-graph.html" . }}
+
+
+
+
+
Network of Contributors
different projects. Each node represents a contributor, and the connections (edges) represent shared project participation. - - + + + diff --git a/scripts/forrt_contribs/tenzing.py b/scripts/forrt_contribs/tenzing.py index ecc642bda9d..06703d476f0 100644 --- a/scripts/forrt_contribs/tenzing.py +++ b/scripts/forrt_contribs/tenzing.py @@ -213,6 +213,14 @@ def fetch_all_contributor_data(df_index): if col in cache_data.columns: cache_data[col] = cache_data[col].astype(str).str.replace(r'[\r\n\t]+', '', regex=True).str.strip() cache_data[col] = cache_data[col].replace('nan', '') + # Standardize names in cache by ORCID: when the same ORCID appears with different + # name variants across projects (e.g. swapped first/surname), use the first + # occurrence as the canonical name. Only affects the cache, not tenzing.md. + name_cols = ['First name', 'Middle name', 'Surname'] + orcid_mask = cache_data['ORCID iD'] != '' + canonical = cache_data.loc[orcid_mask].groupby('ORCID iD')[name_cols].first() + for col in name_cols: + cache_data.loc[orcid_mask, col] = cache_data.loc[orcid_mask, 'ORCID iD'].map(canonical[col]).values cache_data.to_csv(CACHE_FILE, index=False) print(f"💾 Cache saved to {CACHE_FILE} ({len(cache_data)} rows, {len(cache_columns_present)} columns)") diff --git a/scripts/forrt_contribs/tenzing_template.md b/scripts/forrt_contribs/tenzing_template.md index 429520c8b71..14c0a85fd88 100644 --- a/scripts/forrt_contribs/tenzing_template.md +++ b/scripts/forrt_contribs/tenzing_template.md @@ -192,7 +192,7 @@ if (window.location.hash) {
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Network of Contributors
+ +The interactive network visualization below shows the connections +between FORRT contributors based on their collaborative work across +different projects.
+How to read the graph:
+-
+
- Nodes represent contributor names. Each node is +connected to other people involved in the same project(s). +
- Edges (lines) represent connections between +contributors who worked on the same project. +
Use the dropdown menus below to search for a specific contributor by +name or to filter the network by project. Hovering over a node +highlights its connections and shows which projects that person +contributed to. Clicking on a node highlights it in red, and clicking on +an edge highlights the connection between two contributors. You can also +zoom in and out using the scroll wheel or the navigation buttons.
+
+
+
+
+
+
+
+
+
+
+
+