DeepSeekWorkBack officeAI score · 84/100AI-assisted edit · no obvious prompt-injection riskDeepSeek API Docs

Auto-Tag News Text with DeepSeek's Official Content Classification Prompt

Input news text → classify via predefined system prompt → receive single category label

BeginnerStepsPrompt
Who
Content operators or data labelers who need to batch-categorize news, announcements, or social-media content
You get
Classify any Chinese news text into one of: Politics, Economy, Technology, Entertainment, Sports, Education, Health, International, Domestic, Society
You need
DeepSeek

Constraints

Use in any DeepSeek chat interface that supports system/user roles (API or official app). No additional tools or plugins required. Cost depends on the token pricing of the chosen DeepSeek model; not specified in source.

What they did

DeepSeek's official API documentation prompt library provides a 'Content Classification' template. It uses a system prompt to define a 'News Classification Expert' role, limits output to 10 predefined categories, and instructs the model to output only the category name without extra explanation. In the official example, after inputting a news snippet about SpaceX Falcon 9, the model outputs '科技' (Technology).

How it went

The official example shows single-label output '科技', which is an author_report (expected output provided by official docs). Not verified by this site.

Worth knowing

Category schema is fixed to 10 classes; no custom categories or multi-label output. Output format is strictly category name only, no confidence score or explanation. Source does not state support for English or other languages. No API calling code, token pricing details, or batch processing guidance provided.

Steps

  1. Paste the system prompt to set up the classification expert

    Paste the following into the system field (or equivalent) to establish the 'News Classification Expert' role with 10 predefined categories.

    Setup guidance · Source text。DeepSeek conversation system field or equivalent system prompt position

    #### 定位
- 智能助手名称 ：新闻分类专家
- 主要任务 ：对输入的新闻文本进行自动分类，识别其所属的新闻种类。

#### 能力
- 文本分析 ：能够准确分析新闻文本的内容和结构。
- 分类识别 ：根据分析结果，将新闻文本分类到预定义的种类中。

#### 知识储备
- 新闻种类 ：
  - 政治
  - 经济
  - 科技
  - 娱乐
  - 体育
  - 教育
  - 健康
  - 国际
  - 国内
  - 社会

#### 使用说明
- 输入 ：一段新闻文本。
- 输出 ：只输出新闻文本所属的种类，不需要额外解释。

  2. Input text to classify and receive the label

    Enter the news text to classify in a user message; the model will return only a category name without explanation.

    Execution prompt · Source text。User message input; replace with actual text to classify

    美国太空探索技术公司（SpaceX）的猎鹰9号运载火箭（Falcon 9）在经历美国联邦航空管理局（Federal Aviation Administration，FAA）短暂叫停发射后，于当地时间8月31日凌晨重启了发射任务。

Prompt you can copy

Reference prompt · Source text。Confirm the destination, input variables and output format before use.

This block is the prompt. Ads are never placed here.

#### 定位
- 智能助手名称 ：新闻分类专家
- 主要任务 ：对输入的新闻文本进行自动分类，识别其所属的新闻种类。

#### 能力
- 文本分析 ：能够准确分析新闻文本的内容和结构。
- 分类识别 ：根据分析结果，将新闻文本分类到预定义的种类中。

#### 知识储备
- 新闻种类 ：
  - 政治
  - 经济
  - 科技
  - 娱乐
  - 体育
  - 教育
  - 健康
  - 国际
  - 国内
  - 社会

#### 使用说明
- 输入 ：一段新闻文本。
- 输出 ：只输出新闻文本所属的种类，不需要额外解释。

The core reproducible asset is the system prompt text, placed in step 1 prompts. No additional main prompt needed.

Source

DeepSeek 官方提示词：新闻/文本内容自动分类 · DeepSeek · DeepSeek API Docs · 2026-04-17T05:05:00Z

Comments & feedback

Tried this method? Share results, problems or improvements. Comments appear after moderation. Please avoid sensitive information.

Loading…

    Related