Skip to contents

Add network analysis metrics to user profiles

Source: R/algos.R
add_metrics.Rd

Computes and adds comprehensive network analysis metrics to user profile data. This function calculates centrality measures, community detection, and PageRank scores to identify influential users and network structure within your Bluesky social network.

Usage

add_metrics(profiles, net)

Arguments

profiles

Tibble. User profile information from get_profiles containing at least handle and other profile details

net

Tibble. Network edge list with follow relationships containing:

actor_handle

Character. User who is following another

follows_handle

Character. User being followed

Value

Tibble with original profile data enhanced with network metrics:

centrality

Numeric. Betweenness centrality score (influence as bridge)

community

Integer. Community assignment from cluster analysis

pageRank

Numeric. PageRank score (influence based on follower quality)

insideFollowers

Integer. Number of followers within this network

community_label

Character. Descriptive label for the community

Plus all original profile columns (handle, displayName, description, etc.)

Details

The function performs several network analysis computations:

  • Betweenness centrality: Identifies users who act as bridges between communities

  • Community detection: Uses random walks to find densely connected groups

  • PageRank: Measures influence based on follower network quality

  • Inside followers: Counts followers within the analyzed network subset

  • Community labels: Generates descriptive labels using TF-IDF of user bios

See also

build_network, com_labels, create_widget

Other network-analysis: com_labels()

Examples

if (FALSE) { # \dontrun{
# Build a network first
auth <- get_token("your.handle.bsky.social", "your-app-password")
network_result <- build_network(
  key_actors = c("scientist.bsky.social", "researcher.bsky.social"),
  keywords = c("science", "research", "academic"),
  token = auth$accessJwt,
  refresh_tok = auth$refreshJwt,
  threshold = 10
)

# Add comprehensive metrics to profiles
enhanced_profiles <- add_metrics(network_result$profiles, network_result$net)

# Examine the most central (influential) users
top_central <- enhanced_profiles[
  order(enhanced_profiles$centrality, decreasing = TRUE),
][1:10, c("displayName", "handle", "centrality", "pageRank")]
print(top_central)

# Find the most influential users by PageRank
top_pagerank <- enhanced_profiles[
  order(enhanced_profiles$pageRank, decreasing = TRUE),
][1:10, c("displayName", "handle", "pageRank", "followersCount")]
print(top_pagerank)

# Analyze community structure
community_summary <- enhanced_profiles |>
  group_by(community, community_label) |>
  summarise(
    members = n(),
    avg_followers = mean(followersCount, na.rm = TRUE),
    avg_centrality = mean(centrality, na.rm = TRUE),
    .groups = "drop"
  ) |>
  arrange(desc(members))
print(community_summary)

# Find users who bridge communities (high centrality)
bridge_users <- enhanced_profiles[
  enhanced_profiles$centrality > quantile(enhanced_profiles$centrality, 0.9, na.rm = TRUE),
  c("displayName", "handle", "centrality", "community_label")
]
print(bridge_users)

# Compare internal vs external influence
influence_comparison <- enhanced_profiles |>
  select(handle, followersCount, insideFollowers, pageRank) |>
  mutate(
    external_followers = followersCount - insideFollowers,
    influence_ratio = insideFollowers / followersCount
  ) |>
  arrange(desc(influence_ratio))
head(influence_comparison, 10)
} # }