Skip to content

Latest commit

 

History

History
193 lines (141 loc) · 5.89 KB

README.md

File metadata and controls

193 lines (141 loc) · 5.89 KB

Data - Sample Conversation Data

This directory contains sample conversation data files used for testing and demonstration purposes.

📂 Contents

Bilingual Sample Data

All sample data files are now available in both English and Chinese versions:

  • assistant_chat_en.json / assistant_chat_zh.json - Sample assistant-user conversation data

    • One-on-one conversation format
    • User queries and AI assistant responses
    • Used for testing memory extraction from assistant interactions
    • Available in English and Chinese

  • group_chat_en.json / group_chat_zh.json - Sample group conversation data

    • Multi-user group chat format
    • Follows GroupChatFormat specification
    • Used for testing memory extraction from group discussions
    • Available in English and Chinese

Note: The conversation_meta field in the data is provided solely to help users understand the conversation context and participant roles. It is not used during memory extraction and inference generation.

📋 Data Format

GroupChatFormat Specification

All conversation data files follow the standardized GroupChatFormat format:

{
  "version": "1.0.0",
  "conversation_meta": {
    "scene": "group_chat",
    "scene_desc": {},
    "group_id": "group_001",
    "name": "Project Discussion Group",
    "user_details": {
      "user_101": {
        "full_name": "Alice",
        "role": "user",
        "custom_role": "Product Manager"
      }
    }
  },
  "conversation_list": [
    {
      "message_id": "msg_001",
      "create_time": "2025-02-01T10:00:00+00:00",
      "sender": "user_101",
      "content": "Good morning everyone"
    }
  ]
}

Scenario Types

EverMemOS supports two core conversation scenarios:

  • 🤖 Assistant Scenario (scene: "assistant")

    • Human-AI assistant dialogue
    • One-on-one conversation format
    • scene_desc contains description field to describe the conversation scene
    • Example: assistant_chat_en.json, assistant_chat_zh.json

  • 👥 Group Chat Scenario (scene: "group_chat")

    • Multi-person group chat
    • Work collaboration format
    • scene_desc is typically an empty object
    • Example: group_chat_en.json, group_chat_zh.json

📖 Data Scenarios

Group Chat Scenario (group_chat.json)

Background: AI Product Work Group

Project Storylines:

  • Project 1: Add emotion recognition feature to "Smart Sales Assistant"
  • Project 2: Add memory system to "Smart Sales Assistant"

Characters:

  • Alex - AI Algorithm Engineer
  • Betty - Product Manager
  • Chen - Boss
  • Dylan - Backend Engineer
  • Emily - Frontend Engineer

💡 Explore more details using EverMemOS!

Assistant Scenario (assistant_chat.json)

Background: Personal Health & Lifestyle Assistant Conversation

Conversation Topics:

  • Health status consultation
  • Exercise recommendations
  • Dietary preferences tracking
  • Personal profile building

💡 Use EverMemOS to explore our personal conversation data and gain deep insights into how the memory system works!

🎯 Usage

For Demo Scripts

These data files are used by the demo scripts:

# Extract memories from the sample data
uv run python src/bootstrap.py demo/extract_memory.py

The extraction script automatically reads and processes all JSON files in this directory.

For Batch Memory Storage

You can use these files with the batch storage script:

# Validate format (English version)
uv run python src/bootstrap.py src/run_memorize.py \
  --input data/group_chat_en.json \
  --validate-only

# Validate format (Chinese version)
uv run python src/bootstrap.py src/run_memorize.py \
  --input data/group_chat_zh.json \
  --validate-only

# Store to memory system (English version)
uv run python src/bootstrap.py src/run_memorize.py \
  --input data/group_chat_en.json \
  --api-url http://localhost:1995/api/v1/memories

# Store to memory system (Chinese version)
uv run python src/bootstrap.py src/run_memorize.py \
  --input data/group_chat_zh.json \
  --api-url http://localhost:1995/api/v1/memories

📝 Adding Your Own Data

To add your own conversation data:

  1. Create a JSON file following the GroupChatFormat specification
  2. Place it in this directory (data/)
  3. Run validation to ensure format compliance: 
    uv run python src/bootstrap.py src/run_memorize.py \
  --input data/your_data.json \
  --validate-only
  4. Extract memories using the demo or batch scripts

🔍 Data Privacy

Important: The sample data in this directory is for demonstration purposes only and contains fictional conversations. When using EverMemOS with real data:

  • Ensure you have proper consent to process conversation data
  • Follow data privacy regulations (GDPR, CCPA, etc.)
  • Anonymize sensitive information before processing
  • Secure storage of extracted memories

🔗 Related Documentation

📊 Sample Data Statistics

File Messages Users Groups Language Purpose
assistant_chat_en.json 56 2 1 English Assistant conversation demo
assistant_chat_zh.json 56 2 1 Chinese Assistant conversation demo
group_chat_en.json 508 5 1 English Group chat demo
group_chat_zh.json 508 5 1 Chinese Group chat demo

💡 Need Help?

Ready to extract memories! 🧠📊