network_graph() now validates its dcm argument instead of an undefined df, so it no longer fails with object 'df' not found.
association() gained a min_confidence argument. Confidence is now filtered against the confidence threshold rather than the support threshold.
tfidf_vector() now builds one row per article (document) instead of one row per keyword token, so the resulting matrix is a proper document-by-term TF-IDF matrix. Short (2-character) Korean keywords are no longer dropped.
normalize_vector() now performs per-row min-max scaling as documented (previously it applied a single global min-max over the whole matrix).
header_remover() uses a correct character class ([^]]) when stripping bracketed headers.
Data
The bundled sample dataset is now president_lee_2026_08 (6,537 BigKinds articles, August 2026), replacing president_yoon_2023_05. It is stored as a properly named, xz-compressed .rda; the source spreadsheet stays in the repository but is excluded from the build.
Documentation and packaging
DESCRIPTION: fixed the malformed Authors@R entry (email was being parsed as a middle name), dropped the stale hand-written Author/Maintainer fields, added URL, BugReports, and Depends: R (>= 4.2), and rewrote the Description field so it no longer starts with the package name.
Package-level documentation now uses the "_PACKAGE" sentinel.
All files under R/ are now ASCII-only: the Korean BigKinds column names are defined once as \u-escaped constants in R/globals.R and referenced through tidy-evaluation helpers, fixing the “non-ASCII characters” warning.