Modular Toolkit for PageRank Calculation

Provides a set of modular, pipeable functions to calculate PageRank scores from edge lists and redirect reports, common in SEO analysis. Functions handle URL cleaning, redirect resolution, edge deduplication, isolate handling, and PageRank computation using base R for data manipulation and 'igraph' for core PageRank calculation.

pagerankr: SEO-Focused PageRank Modeling Toolkit

• 1 pagerankr
  • 1.1 Installation
  • 1.2 Quick Start
  • 1.3 Current Capabilities
    • 1.3.1 Crawl Data Prep and Redirect QA
    • 1.3.2 Screaming Frog Crawl Imports
    • 1.3.3 PageRank Modeling Controls
    • 1.3.4 Comparison, Grid Search, and Simulation
    • 1.3.5 Export and Exploration
  • 1.4 Key Functions
  • 1.5 Further Information
  • 1.6 Code of Conduct
  • 1.7 License

1 pagerankr

pagerankr is an SEO-focused R toolkit for PageRank modeling on crawl data. It supports both a single end-to-end wrapper (pagerank()) and modular building blocks for cleaning URLs, auditing redirects, resolving link graphs, running scenario comparisons, and exporting graph outputs.

The package currently includes:

• End-to-end and low-level PageRank calculation
• Redirect diagnostics and resolution policies
• Nofollow and indexability-aware PageRank modeling
• Domain/host filtering and edge-weight transformations
• Screaming Frog Internal: All plus All Inlinks/Outlinks import adapters
• Model comparison, parameter sweeps, and what-if simulations
• Graph export utilities and an optional interactive Shiny explorer
• HITS hub and authority scores (hits())
• SALSA hub and authority scores (salsa())
• TrustRank seed-biased PageRank (trustrank())
• Topic-Sensitive PageRank (per-topic authority with blended scores)
• Reverse-graph feeder PageRank (find pages that power a cluster)
• GA4 behavioral transition modeling (transition counts, structural smoothing, entrance teleport)
• PageRank convergence controls (algo/eps/niter) and alpha-stability reporting (pagerank_stability())

1.1 Installation

# From CRAN
install.packages("pagerankr")

# Development version from GitLab
# install.packages("devtools")
devtools::install_gitlab("bart-turczynski/pagerankr")

1.2 Quick Start

library(pagerankr)

edges <- data.frame(
  from = c("http://example.com/home",
           "http://example.com/about",
           "http://example.com/blog"),
  to   = c("http://example.com/about",
           "http://example.com/home",
           "http://example.com/home")
)

redirects <- data.frame(
  from = "http://example.com/old-blog",
  to   = "http://example.com/blog"
)

pr <- pagerank(edges, redirects_df = redirects)
print(pr)

1.3 Current Capabilities

1.3.1 Crawl Data Prep and Redirect QA

• clean_url_columns() canonicalizes URL columns using rurl::get_clean_url
• audit_redirects() reports redirect chains, loops, conflicts, self-refs, and optional orphaned rules vs. an edge list
• resolve_redirects() applies redirect maps to an edge list with conflict and loop policies
• resolve_redirect_urls() resolves a character vector of URLs without requiring an edge list
• resolve_links() returns the resolved/deduplicated graph without computing PR
• get_unique_edges() and drop_isolates() provide explicit graph hygiene tools

audit <- audit_redirects(redirects, edge_list_df = edges)
print(audit)

resolve_redirect_urls(
  c("http://example.com/old-blog", "http://example.com/home"),
  redirects
)

1.3.2 Screaming Frog Crawl Imports

Use screaming_frog_bundle() with an Internal: All export and either All Inlinks or All Outlinks. The node export supplies page facts, redirects, canonicals, and indexability. The link export supplies raw link observations and graph-eligible Hyperlink edges. Resource, canonical, hreflang, and other non-Hyperlink link rows are retained in diagnostics but excluded from the PageRank graph by default.

bundle <- screaming_frog_bundle(
  internal = "internal_all.csv",
  links = "all_outlinks.csv",
  link_export_kind = "all_outlinks"
)

pr <- pagerank_screaming_frog(bundle)
attr(pr, "screaming_frog_import")
attr(pr, "transition_audit")

Placement and rendered-vs-HTML policies are explicit scoring choices:

pagerank_screaming_frog(
  bundle,
  accepted_placements = c("nav", "content"),
  link_origins = c("html", "html_rendered"),
  placement_weights = c(nav = 2, content = 1)
)

1.3.3 PageRank Modeling Controls

• pagerank() supports weighted edges via weight_col
• duplicate_edge_policy = "collapse" keeps the standard binary destination-level surfer as the default: repeated from -> to rows become one edge. Opt into "aggregate" to sum duplicate numeric weights, or "count_instances" for a link-slot surfer where repeated links to the same target increase transition probability.
• nofollow_col + nofollow_action = c("evaporate", "drop", "keep")
• indexability_df support for noindex and Blocked by robots.txt behaviors (robots_blocked_action = "show" or "vanish")
• Domain scoping directly in pagerank() (keep_domains, exclude_domains) or via filter_links_by_domain() with domain/host keep/ignore rules
• transform_weights() provides rank/log/zipf/percentile transforms for raw edge signals

edges_w <- data.frame(
  from = c("Home", "Home", "Home"),
  to = c("About", "Blog", "Contact"),
  position = c(1, 2, 5)
)

edges_w$weight <- transform_weights(
  edges_w$position,
  method = "zipf",
  descending = FALSE
)

pagerank(edges_w, weight_col = "weight", clean_edge_urls = FALSE)

1.3.4 Comparison, Grid Search, and Simulation

• compare_pagerank() calculates deltas, rank shifts, and summary stats
• auto_grid() and pagerank_grid() run parameter sweeps
• analyze_pagerank_grid() summarizes concentration/distribution effects
• simulate_changes() compares baseline vs proposed links/redirects
• pr_gini(), pr_entropy(), and pr_top_k_share() compute distribution metrics

grid <- auto_grid(
  damping = c(0.85, 0.95),
  nofollow_action = c("evaporate", "drop")
)

grid_results <- pagerank_grid(edges, params_grid = grid, clean_edge_urls = FALSE)
analyze_pagerank_grid(grid_results)

1.3.5 Export and Exploration

• export_graph() writes outputs in graphml, dot, edgelist, or pajek formats
• launch_pagerank_explorer() launches an interactive Shiny app for uploads, visualization, redirect auditing, and exports

pr <- pagerank(edges, clean_edge_urls = FALSE)
export_graph(pr, edges, file = "pagerank.graphml", format = "graphml")

# Optional interactive app:
# install.packages(c("shiny", "DT", "visNetwork"))
# launch_pagerank_explorer()

1.4 Key Functions

1.5 Further Information

The reference and vignettes ship with the package and are reachable through help(package = "pagerankr") and the vignette() calls below; the rendered website is being rebuilt after the move to GitLab. To report a bug or request an enhancement, use GitLab Issues. Please read CONTRIBUTING.md before proposing a change; it sets out the test, lint, and R CMD check requirements. For privately reported security vulnerabilities, follow SECURITY.md.

help(package = "pagerankr")
vignette("pagerankr-usage")
vignette("trustrank")
vignette("topic_feeder_pagerank")

1.6 Code of Conduct

Please note that the pagerankr project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.

1.7 License

MIT License. See the LICENSE file for details.