SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsdomain-classifier

domain-classifier

Design & creation

View repository

Updated 10-15 22:14

About this skill

Domain Classifier - AI Domain Classifier

Version: 1.0 Purpose: Intelligently analyzes prompt content and accurately determines the domain it belongs to Output: Structured JSON format

🎯 Task

Analyze the prompt content and determine which of the following 7 major domains it belongs to:

Definitions of the 7 Major Domains

1. portrait (Portrait Photography)

  • Character portraits, facial features, expressions, poses
  • Makeup, hairstyle, clothing, accessories
  • Portrait photography techniques, lighting, angles
  • Keywords: person, face, eyes, skin tone, makeup, hairstyle, expression, pose

2. interior (Interior Design)

  • Interior spaces, furniture layout, decoration styles
  • Living room, bedroom, kitchen, office, and other spaces
  • Material combinations, spatial atmosphere
  • Keywords: interior, room, furniture, space, decoration, layout

3. product (Product Photography)

  • Product display, commercial photography
  • Items, goods, packaging, still life
  • Product close-ups, material display
  • Keywords: product, item, goods, packaging, bottle, watch, bag

4. design (Graphic Design)

  • Posters, UI interfaces, infographics, business cards
  • Typography, grid layout, Bento Grid
  • Graphic design, visual communication
  • Keywords: poster, design, layout, grid, UI, typography, graphics

5. art (Art Style)

  • Painting, illustration, artistic creation
  • Ink wash, oil painting, watercolor, sketch
  • Art styles, schools, techniques
  • Keywords: painting, art, ink wash, oil painting, illustration, artwork

6. video (Video Generation)

  • Video scenes, camera movement
  • Dynamic effects, storyboards, action
  • Movie scenes, animation
  • Keywords: video, camera, movement, storyboard, movie, animation, scene transition

7. common (General Photography)

  • Photography techniques, camera parameters
  • Lighting, composition, depth of field
  • Photography techniques not specific to a particular domain
  • Keywords: photography, camera, lens, lighting, focus, depth of field, composition

📋 Analysis Process

Step 1: Read the prompt

Carefully read the complete content of the prompt and understand its core theme and purpose.

Step 2: Identify key elements

Identify from the prompt:

  • Subject: What is being described (person/object/space/design)?
  • Visual style: Is it photography/painting/design/motion?
  • Technical features: Which domain-specific terms are used?
  • Application scenario: What is it used for?

Step 3: Domain matching

Based on the key elements, determine the best-matching domain:

Priority rules：

  1. If the subject isa person's face/full body → portrait
  2. If the subject isan interior space/furniture → interior
  3. If the subject isproduct itemand emphasizes commercial display → product
  4. if it isposter/UI/graphic work → design
  5. if it ispainting/artistic creation → art
  6. if it emphasizesvideo/motion/camera movement → video
  7. if it is only generalphotography technique → common

Special case handling：

  • if it includes people but the focus isspace → interior
  • if it includes a product but the focus isdesign layout → design
  • if it is a person but usespainting style → art
  • if it cannot be clearly determined → choose the best-matching field

Step 4: Calculate confidence

Evaluate confidence (0-100%) based on the following factors:

  • 90-100%: the field features are very clear, including multiple keywords for that field
  • 70-89%: the field features are clear, with obvious markers
  • 50-69%: the field features are basically clear, but somewhat vague
  • 30-49%: the field features are not obvious, and it may belong to multiple fields
  • 0-29%: the field cannot be determined

📤 Output format

mustoutput in JSON format:

{
  "primary_domain": "portrait",
  "confidence": 85,
  "reasoning": "提示词描述了一位女性的面部特征，包括眼睛、肤色、妆容等人像摄影元素，明确属于人像摄影领域。",
  "secondary_domains": ["common"],
  "keywords_found": ["面部", "肤色", "妆容", "摄影"]
}

Field description：

  • primary_domain: primary field ID (portrait/interior/product/design/art/video/common)
  • confidence: confidence (0-100 integer)
  • reasoning: reason for the judgment (1-2 sentences, Chinese)
  • secondary_domains: secondary related fields (optional, array)
  • keywords_found: keywords found (optional, array)

🔍 Example analysis

Example 1: Portrait photography

Input：

一位20岁的亚洲女性，长发飘逸，淡妆，温柔的微笑，柔和的工作室光照，
浅景深，背景虚化，85mm镜头，f/1.8光圈。

Output：

{
  "primary_domain": "portrait",
  "confidence": 95,
  "reasoning": "明确描述人物肖像，包含年龄、性别、面部特征、表情等人像关键元素，且使用人像摄影技术。",
  "secondary_domains": ["common"],
  "keywords_found": ["女性", "面部", "微笑", "工作室光照", "浅景深", "85mm镜头"]
}

Example 2: Interior design

Input：

现代简约风格客厅，米色沙发，木质茶几，落地窗，自然光照，
极简主义，温馨氛围，3D渲染。

Output：

{
  "primary_domain": "interior",
  "confidence": 92,
  "reasoning": "描述室内空间布局和家具，明确是客厅场景，属于室内设计领域。",
  "secondary_domains": [],
  "keywords_found": ["客厅", "沙发", "茶几", "室内", "空间"]
}

Example 3: Graphic design

Input：

Bento网格布局海报，8个模块，非对称排列，玻璃态效果，
现代排版，信息图表，横向16:9。

Output：

{
  "primary_domain": "design",
  "confidence": 95,
  "reasoning": "明确的平面设计作品，包含Bento网格、海报、排版等设计关键词。",
  "secondary_domains": [],
  "keywords_found": ["Bento网格", "海报", "布局", "排版", "信息图"]
}

Example 4: Art style

Input：

中国传统水墨画，山水意境，泼墨技法，写意风格，
宣纸质感，黑白灰层次，留白艺术。

Output：

{
  "primary_domain": "art",
  "confidence": 98,
  "reasoning": "典型的艺术绘画作品，使用传统绘画技法和艺术表现手法。",
  "secondary_domains": [],
  "keywords_found": ["水墨画", "绘画", "技法", "艺术"]
}

Example 5: Ambiguous case (product vs design)

Input：

Evian矿泉水瓶产品展示，Bento网格布局，8个信息卡片，
玻璃材质，产品信息图表设计。

Output：

{
  "primary_domain": "design",
  "confidence": 65,
  "reasoning": "虽然包含产品（Evian瓶），但重点是Bento网格布局和信息设计，属于平面设计范畴。",
  "secondary_domains": ["product"],
  "keywords_found": ["Bento网格", "布局", "信息图表", "设计", "产品"]
}

⚠️ Important rules

  1. Must choose one primary field - cannot return None or null
  2. Confidence cannot be lower than 30% - even if uncertain, make the best judgment
  3. reasoning must be specific - explain why this field was chosen
  4. Prioritize the subject object - first look at what is described, then look at technique and style
  5. Chinese first - use Chinese when outputting reasoning and keywords_found

🎯 Batch processing mode

When multiple prompts need to be analyzed in batches, output a JSON array:

[
  {
    "file_id": "prompt_001.txt",
    "primary_domain": "portrait",
    "confidence": 85,
    "reasoning": "..."
  },
  {
    "file_id": "prompt_002.txt",
    "primary_domain": "design",
    "confidence": 72,
    "reasoning": "..."
  }
]

When using this Skill, strictly follow the output format and ensure the JSON is valid!