You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Dataset Card for tw-science-24M

本資料集收錄臺灣相關之科學/科技類繁體中文文本(涵蓋農業、生態、氣候、醫療、化學、生物等領域),總 token 數約 24M(24 百萬),可作為繁中模型在科學領域的補充預訓練語料。

Dataset Details

Dataset Description

資料來自繁體中文公開科學/科技/環境議題之報導與政府公告,內容包含:

  • 科學新聞、政策報導
  • 永續、生態保育、氣候議題
  • 醫療衛生、農業科技
  • 部分 zh-en 平行段落

文本經過清理、保留段落結構,可直接用於語言模型的持續預訓練。

  • Curated by: Huang Liang Hsun
  • Language(s) (NLP): Traditional Chinese, English
  • License: cc-by-nc-sa-4.0

Dataset Sources

Uses

Direct Use

  • 繁中科學語料補充:補強模型對臺灣科技政策、環境議題的覆蓋。
  • 衍生科學 chatbot、永續報告 RAG 系統的知識來源。

Out-of-Scope Use

  • 不適用作為醫療診斷、藥物建議;資料為一般報導。
  • 不適合作為主訓練語料;建議作為 sub-mix。

Dataset Structure

{
  "text": "(科學/環境/永續類繁中段落樣本)..."
}

僅含單一欄位 text

Dataset Creation

Curation Rationale

通用繁中語料雖大,但科學/科技領域的臺灣本土語料相對稀缺。本資料集針對該領域做集中蒐集,補強模型對科學/永續/環境議題的覆蓋。

Source Data

Data Collection and Processing

  1. 蒐集臺灣公開科學/科技/環境媒體與政府公告。
  2. 移除 HTML 雜訊,合併段落為純文字。
  3. 計算累積 token 數,輸出至約 24M tokens 的目標規模。

Who are the source data producers?

原始文章由各媒體編輯與政府機關公開發布。

Annotations

Annotation process

無人工標註。

Who are the annotators?

無標註者。

Personal and Sensitive Information

資料為公開報導/公告,所提及之人名屬公眾人物身份。

Bias, Risks, and Limitations

  • 報導取樣帶有媒體編輯選題偏好,部分議題覆蓋密度不均。
  • 內容反映報導當時的政策/科學共識,可能與最新狀況有時差。

Recommendations

建議與其他繁中通用語料一同混訓並控制比重;下游若用於專業領域應用,請結合最新權威來源。

Citation

@misc{tw_science_24m,
  title        = {tw-science-24M: Traditional Chinese Science and Technology Corpus from Taiwan (24M tokens)},
  author       = {Huang, Liang Hsun},
  year         = {2024},
  howpublished = {\url{https://huggingface.co/datasets/lianghsun/tw-science-24M}}
}

Dataset Card Authors

Huang Liang Hsun

Dataset Card Contact

Huang Liang Hsun

Downloads last month
14

Models trained or fine-tuned on lianghsun/tw-science-24M