Changes in version 1.0.0 (2026-06-08)                  

New features

  - First release as an R package built with golem.
  - Five-step pipeline: Import → Preprocess → Inference → Final Model →
    Trends.
  - Multi-file upload: multiple Scopus CSVs and Web of Science TXTs can
    be merged in a single operation with DOI-based deduplication.
  - Custom ggplot2 plots with full export control (PNG / TIFF / JPEG /
    PDF).
  - Word cloud per topic with RColorBrewer palette support.
  - Topic trend classification (HOT / COLD / EQUAL) via linear
    regression.

Breaking changes

None (first release).

Bug fixes

  - options(shiny.maxRequestSize) is now set and restored within
    run_LDAShiny() only — not at package load time.
  - Removed audio dependency (beepr) which is unsuitable for CRAN
    packages.
  - Fixed get(input$theme_choice) calls replaced with safe named-list
    lookups using lda_themes().
  - mod_inference_server is now properly exported.
  - parallel::detectCores() is now declared in NAMESPACE via
    @importFrom.