rouge-chinese

What it is and what it does

Rouge-Chinese is a Python library that calculates ROUGE evaluation scores specifically for Chinese-language NLP tasks. It addresses known limitations in the original ROUGE implementation when applied to Chinese text: incorrect sentence segmentation (missing Chinese punctuation marks), excessive memory consumption during longest-common-subsequence calculation, and inaccurate scores due to approximation methods. The library improves sentence splitting to recognize Chinese punctuation, optimizes memory usage by computing sequence lengths without generating the sequences themselves, and computes official ROUGE scores rather than approximate variants.

The package provides three main interfaces: a library API for scoring single or multiple sentence pairs with optional averaging, a file-based API for batch scoring line-by-line from text files, and a command-line tool for ad-hoc scoring. It depends only on six and requires external word segmentation to tokenize Chinese text before scoring. Output includes precision, recall, and F1 scores for ROUGE-1, ROUGE-2, and ROUGE-L metrics.

Use it for:

• Evaluate machine-generated Chinese text summaries against reference summaries in NLP research or production systems.
• Batch-score multiple hypothesis-reference pairs from files to assess summarization model performance across datasets.
• Compare Chinese abstractive or extractive summarization outputs in academic papers or model benchmarking.
• Integrate into Chinese NLP pipelines where ROUGE metrics are required for quality assurance or model selection.
• Command-line scoring of individual Chinese text pairs without writing Python code.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Computes ROUGE evaluation metrics for Chinese text summarization and NLP tasks, handling Chinese punctuation and sentence segmentation correctly.

Yes, if you need ROUGE metrics for Chinese text and can tolerate dormant maintenance. The package solves real problems in the original ROUGE for Chinese, has low install friction, and no known vulnerabilities. However, verify that the unclear license aligns with your use case, and confirm compatibility with your Python version since support is unspecified. Consider it stable for evaluation tasks but not actively developed.

Install

rouge-chinese on PyPI

pip install rouge-chinese

uv add rouge-chinese

poetry add rouge-chinese

Installing rouge-chinese

Before you install

Low friction: pure Python wheel with a single runtime dependency (six). Dormant since first release with no updates, though the repository remains active with 114 stars.

License in practice

License treatment is unclear—the package references LICENCE.txt but provides no SPDX identifier. Review the license file in the repository before adopting in proprietary or restricted contexts.

Quickstart

pip install rouge-chinese

from rouge_chinese import Rouge

rouge = Rouge()
scores = rouge.get_scores(hypothesis, reference)

Requires external word segmentation to tokenize Chinese text before scoring; the package itself does not include tokenization.

Verify before relying

• Whether the package works correctly with modern Python versions (python_support is unspecified in metadata)
• Current state of the single runtime dependency (six) and its long-term maintenance status
• Whether memory optimizations claimed in the description remain effective with contemporary hardware and text sizes

Package facts

Evidence: rouge_chinese-1.0.3-py3-none-any.whl

Keywords: NL, CL, natural language processing, computational linguistics, summarization, chinese

Tags

More Linguistic packages