Skip to contents

Changelog

Source: NEWS.md

jiebaRS (development version)

  • Fixed source installation on native Windows ARM64 by selecting the matching Rust target.
  • Added import_cidian() to import words from SCEL, QCEL, QPYD, BDICT, and BCD input-method dictionaries into an existing worker.
  • Added read_cidian() to read supported input-method dictionaries into a data frame of entries and coding components.
  • Bundled stopwords_cn, stopwords_en, and stopwords_full stopwords datasets are now available for explicit use; stopword filtering remains disabled by default.
  • Removed get_tuple(); use count_ngrams() for n-gram counting.
  • Exported functions now reject unnamed optional arguments; supply optional arguments by name.
  • Documentation for internal native bridge functions is no longer generated for users.

jiebaRS 0.2.0

CRAN release: 2026-08-04

  • Custom dictionary and model files are now read through a common UTF-8 layer that handles a leading BOM and reports invalid formats.
  • Reject duplicate entries in custom IDF dictionaries now.
  • User dictionary entries with an omitted frequency now infer one automatically, zero frequencies are rejected.
  • Legacy word tag entries from jiebaR are supported.
  • worker() now accepts min_keyword_length to control the minimum Unicode length of terms returned by TF-IDF and TextRank keyword extraction.
  • worker() now accepts one or more user dictionary paths through user; dictionaries are appended in the supplied order (qinwf/jiebaR#69).

jiebaRS 0.1.0

Initial CRAN submission.

Implemented these APIs:

  • Workers:
    • worker
  • Segmentation:
    • segment
    • segment_batch
  • Speech Tagging:
    • tagging
    • tagging_batch
  • Keyword Extraction:
    • keywords
    • keywords_df
    • textrank
    • textrank_df
  • Word Frequency and N-grams:
    • freq
    • count_ngrams
    • get_tuple
  • Utilities:
    • filter_segment
    • new_user_word
    • add_word
    • get_idf

Add necessary tests, documents, a benchmark, and the website.