亿韵启界 E_rhythm

E_rhythm project deeply integrates advanced AI technology and data analytics into traditional Chinese poetry creation, establishing a modern platform that combines creation, sharing, and interaction. This platform provides users with a novel poetic experience and drives the digital dissemination of traditional Chinese culture.

Program structure

*Only crucial parts are listed down below

• Main Module - Primary Node.js Web Application

  • routes - Express.js route handlers
  • models - Data access layers
  • views - EJS template files with assets
  • app.js - Express application entry point

• MixPoet Module - AI Poetry Generation Module

  • checkpoint - Trained model weights [download link]

  • codes - Core poetry generation source code

    • app.py - MixPoet application entry point

  • corpus - Poetry corpus and vocabulary

  • data - Additional poetry data resources

  • preprocess - Data preprocessing utilities

• Data Analysis Module - Flask Data Analysis Platform

  • app.py - Dash application main file

• While running, Main Module is on port 3000, MixPoet Module on port 5000, Data Analysis Module on port 8050.

Database Schema

• User

  user
  ├── id (int)                     # Primary key
  ├── username (varchar(200))      # Username
  ├── email (varchar(50))          # User email
  ├── password (varchar(200))      # Encrypted password
  ├── regtime (datetime)           # Registration timestamp
  ├── goodpostNum (int)            # Number of quality posts
  ├── introduction (varchar(300))  # Personal introduction
  ├── contribution (int)           # Contribution value
  ├── post_num (int)               # Total post count
  └── avatar (varchar(255))        # Avatar image path
  

• Post

  post
  ├── id (int)                     # Primary key
  ├── uid (int)                    # Foreign key to user.id
  ├── content (varchar(1000))      # Post content
  ├── title (varchar(100))         # Post title
  ├── posttime (datetime)          # Publication timestamp
  ├── like (int)                   # Like status flag
  ├── collection (int)             # Collection status flag
  ├── type (int)                   # Post type
  ├── like_count (int)             # Likes counter
  └── collection_count (int)       # Collections counter
  

• User Post Actions

  user_post_actions
  ├── id (int, auto increment)     # Primary key
  ├── post_id (int)                # Foreign key to post.id
  ├── user_id (int)                # Foreign key to user.id
  ├── is_liked (tinyint(1))        # Like status
  ├── is_collected (tinyint(1))    # Collection status
  └── created_at (timestamp)       # Action timestamp, defaults to current time
  

• Reply

  reply
  ├── id (int)                     # Primary key
  ├── pid (int)                    # Foreign key to post.id
  ├── uid (int)                    # Foreign key to user.id
  ├── content (varchar(1000))      # Reply content
  └── posttime (datetime)          # Reply timestamp
  

• Poetry Imagery Encyclopedia

  yixiang
  ├── id (int, auto increment)     # Primary key
  ├── name (varchar(50))           # Name
  ├── body (varchar(100))          # Essence of imagery
  ├── meaning (varchar(255))       # Meaning explanation
  ├── example (text)               # Usage examples
  └── category (varchar(50))       # Category
  

Acknowledgements

This project incorporates MixPoet, an AI poetry generation framework developed by the THUNLP-AIPoet team at Tsinghua University, and we trained the model on our own. We express our sincere gratitude to the researchers for making this exceptional work publicly available.

References:

• Repository: https://github.com/THUNLP-AIPoet/MixPoet

• Citation: Xiaoyuan Yi, Ruoyu Li, Cheng Yang, Wenhao Li and Maosong Sun. 2020. MixPoet: Diverse Poetry Generation via Learning Controllable Mixed Latent Space. In Proceedings of The Thirty-Fourth AAAI Conference on Artificial Intelligence, New York, USA.