Internal constructor used by [pagerank()] to assemble the audit record from counts gathered along the aggregation / validation / cleaning path. Every argument has a default so that partially-known states (e.g. an empty edge list) still produce a well-formed object with the documented fields present.
Usage
new_transition_audit(
n_input_rows = 0L,
n_edges = 0L,
n_vertices = 0L,
weighted = FALSE,
weight_col = NULL,
n_edges_weighted = 0L,
duplicate_edge_policy = "collapse",
instance_count_col = NULL,
n_duplicate_instances = 0L,
duplicate_edges = NULL,
n_rows_na = 0L,
n_rows_duplicate = 0L,
n_self_loops = 0L,
n_prior_unmatched = NA_integer_,
n_robots_blocked = 0L,
pagerank_total = NA_real_,
mass_reported = NA_real_,
mass_evaporated = NA_real_,
mass_leaked = NA_real_,
mass_hidden = NA_real_,
out_of_scope_fold = "relabel",
n_out_of_scope_folds = 0L,
out_of_scope_folds_applied = TRUE,
out_of_scope_fold_list = NULL,
fold_collisions = NULL,
config = list()
)Arguments
- n_input_rows
Integer, rows in the raw `edge_list_df`.
- n_edges
Integer, directed edges that survived folding, dedup and self-loop handling (the edges actually scored).
- n_vertices
Integer, vertices in the returned result.
- weighted
Logical, whether an edge `weight_col` was in effect.
- weight_col
Character or `NULL`, the weight column name.
- n_edges_weighted
Integer, edges carrying a finite positive weight.
- duplicate_edge_policy
Character, the duplicate-edge policy used by [pagerank()].
- instance_count_col
Character or `NULL`, internal count column used by `duplicate_edge_policy = "count_instances"`.
- n_duplicate_instances
Integer, duplicate link instances folded into transition weights.
- duplicate_edges
Data frame or `NULL`, compact counted-edge audit rows.
- n_rows_na
Integer, input rows dropped due to `NA` endpoints.
- n_rows_duplicate
Integer, rows collapsed by deduplication.
- n_self_loops
Integer, self-loop edges dropped.
- n_prior_unmatched
Integer or `NA`, prior URLs that did not fold onto a vertex.
- n_robots_blocked
Integer, URLs treated as robots.txt-blocked.
- pagerank_total
Numeric, sum of the returned PageRank scores.
- mass_reported
Numeric, stationary mass on returned/visible pages (typically equal to `pagerank_total`).
- mass_evaporated
Numeric, stationary mass sent to the nofollow evaporation sink (authority wasted on nofollowed outlinks). `0` when no evaporation occurred.
- mass_leaked
Numeric, stationary mass sent to the leak sink under `out_of_scope_fold = "leak"` (authority that flowed into out-of-scope-folded sources, treated like an external redirect). `0` when no leak occurred.
Numeric, stationary mass trapped on hidden / vanished robots-blocked nodes that were removed from the results. `0` when none.
- out_of_scope_fold
Character, the `out_of_scope_fold` policy used (`"relabel"`, `"keep"` or `"leak"`).
- n_out_of_scope_folds
Integer, count of composed fold-map entries whose target was not a crawled node.
- out_of_scope_folds_applied
Logical, `TRUE` when the out-of-scope folds were acted upon (relabeled under `"relabel"`, or routed to the leak sink under `"leak"`), `FALSE` when they were skipped (kept) under `"keep"`.
- out_of_scope_fold_list
Data frame or `NULL`, the out-of-scope folds as `source` / `target` / `signal` rows.
- fold_collisions
Data frame or `NULL`, fold-target collisions detected on the pre-fold edge list: rows of `target` / `n_independent_refs` / `source` for uncrawled URLs that a fold relabeled a crawled source onto while they were also independently linked. `NULL` when no `indexability_df` crawl-URL set was available to detect them.
- config
A named list of the relevant [pagerank()] configuration.
Examples
# Low-level plumbing: normally you obtain a transition_audit via
# attr(pagerank(...), "transition_audit") rather than by hand. Every
# argument defaults, so a bare call yields a well-formed, empty-graph object.
audit <- new_transition_audit()
audit$counts
#> $n_input_rows
#> [1] 0
#>
#> $n_edges
#> [1] 0
#>
#> $n_vertices
#> [1] 0
#>
# Populate a few fields to describe a small scored graph.
audit <- new_transition_audit(
n_input_rows = 4L,
n_edges = 3L,
n_vertices = 3L,
n_rows_duplicate = 1L,
pagerank_total = 1,
mass_reported = 1
)
audit$counts$n_edges
#> [1] 3
audit$dropped$n_rows_collapsed
#> [1] 1