pagerankr: SEO-Focused PageRank Modeling Toolkit

1 pagerankr

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

# 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

Function	Purpose
`pagerank()`	End-to-end PageRank pipeline
`compute_pagerank()`	Low-level wrapper around `igraph::page_rank()`
`resolve_links()`	Resolve redirects and deduplicate graph without PR
`resolve_redirects()`	Apply redirect rules to an edge list
`resolve_redirect_urls()`	Resolve standalone URL vectors through redirects
`resolve_canonicals()`	Apply rel=canonical folds to edge endpoints
`resolve_folded_urls()`	Resolve URL vectors through redirects plus canonicals
`audit_redirects()`	Diagnose redirect chains, loops, and conflicts
`screaming_frog_bundle()`	Compose Screaming Frog node and link exports
`pagerank_screaming_frog()`	Score a Screaming Frog bundle via `pagerank()`
`clean_url_columns()`	Canonicalize URL columns in data frames
`get_unique_edges()`	Deduplicate edges and handle self-loops
`drop_isolates()`	Build vertex sets with or without isolates
`filter_links_by_domain()`	Filter edges by keep/ignore domain or host lists
`transform_weights()`	Transform raw signals into PageRank edge weights
`compare_pagerank()`	Compare two PageRank outputs with rank deltas
`simulate_changes()`	Evaluate proposed link/redirect changes
`auto_grid()`	Build exhaustive parameter grids
`pagerank_grid()`	Run PageRank across multiple parameter sets
`analyze_pagerank_grid()`	Summarise PageRank distribution by model
`pr_gini()`	Gini concentration metric
`pr_entropy()`	Entropy dispersion metric
`pr_top_k_share()`	Top-k PageRank concentration share
`export_graph()`	Export graph and PageRank metadata for external tools
`launch_pagerank_explorer()`	Start the interactive Shiny explorer
`hits()`	End-to-end HITS hub + authority scores
`compute_hits()`	Low-level igraph HITS wrapper
`salsa()`	End-to-end SALSA hub + authority scores
`compute_salsa()`	Low-level SALSA computational core
`trustrank()`	TrustRank: seed-biased PageRank from a trusted seed set
`topic_sensitive_pagerank()`	Per-topic personalized PageRank with blended scores
`topic_feeder_pagerank()`	Reverse-graph seeded PR: find pages that feed a cluster
`seed_prior()`	Build a teleport prior from a seed set (for trustrank / topic_feeder_pagerank)
`align_prior_to_vertices()`	Align a prior/teleport data frame to the graph vertex set
`damping_sensitivity()`	Sweep PageRank across a range of damping factors
`pagerank_stability()`	Alpha-stability report: rank correlation across a damping grid
`ga4_page_transitions()`	Consecutive page-view transition counts from a GA4 export
`smooth_transitions()`	Shrink sparse empirical transitions toward a structural prior
`ga4_entrance_teleport()`	Entrance/landing-page counts as a PageRank teleport vector
`aggregate_edges()`	Aggregate duplicate edges after URL folding
`transform_edge_weights()`	Per-source grouped edge weight transforms
`validate_edge_weights()`	Validate per-source weight totals
`screaming_frog_internal()`	Import Screaming Frog Internal: All export
`screaming_frog_links()`	Import Screaming Frog All Inlinks / All Outlinks export
`audit_redirects()`	Diagnose redirect chains, loops, and conflicts
`audit_canonicals()`	Diagnose rel=canonical fold coverage and conflicts
`resolve_canonicals()`	Apply rel=canonical folds to an edge list
`resolve_canonical_urls()`	Resolve a URL vector through rel=canonical folds
`resolve_folded_urls()`	Resolve a URL vector through redirects plus canonicals

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.