SIU director's-reports corpus: reviewed data first, fetch only what's new

The right way to get SIU data in the morie ecosystem. Loads the panel-reviewed 65-column corpus bundled in rmoriedata (2,182 English reports, subject-official coverage 100 percent, built by a multi-model reading panel plus deterministic residual resolution) – nothing is re-fetched or re-parsed for reports already reviewed. With update = TRUE it then discovers reports published AFTER the corpus snapshot (report ids above the bundled maximum), fetches only those few (through rmoriebricklayer's live-plus-Wayback engine when installed), fills the mechanical schema fields with the compiled parser, and appends them flagged panel_reviewed = FALSE with the judgment columns left NA – run the reading panel (the siu command-line tool against your own 'Ollama' server) to fill those, exactly as the reviewed corpus was built.

Usage

morie_siu_reports(update = FALSE, max_new = 25L, quiet = FALSE)

Arguments

update: Also fetch + parse reports newer than the bundled corpus (default FALSE: fully offline).
max_new: Ceiling on how many new reports to fetch per call (default 25; a normal refresh sees 0-15).
quiet: Suppress progress messages.

Value

A data.frame in the 65-column reviewed-corpus schema. New rows (if any) carry panel_reviewed = FALSE.

Examples

df <- morie_siu_reports()
nrow(df)
#> [1] 5157
table(df$panel_reviewed)
#> 
#> FALSE  TRUE 
#>  2975  2182