Skip to content

Latest commit

 

History

History
116 lines (87 loc) · 3.7 KB

README.md

File metadata and controls

116 lines (87 loc) · 3.7 KB

🍙 Bab-Al 🍙

AI-based Healthy Eating Habits Management Platform


로고

2024.03.06 ~ 2024.11.29


🔽🔽🔽
Redirect to Main README


🎀 Architecture


🕹️ Tech Stack

  • Core Dependencies

    • spring-boot-starter-data-jpa
    • spring-boot-starter-web
    • spring-boot-starter-test
    • spring-boot-starter-validation
    • lombok
  • Database

    • mysql-connector-java:8.0.33
  • Security & Authentication

    • spring-boot-starter-security
    • jjwt-api:0.11.5
    • jjwt-impl:0.11.5
    • jjwt-jackson:0.11.5
  • API Documentation

    • springdoc-openapi-starter-webmvc-ui:2.0.2
  • Email

    • spring-boot-starter-mail
  • Entity Management

    • javassist:3.15.0-GA
  • External Libraries

    • opencsv:5.9
    • jython-slim:2.7.4rc1

🎭 TroubleShooting

🍽️ Improve accuracy and increase diversity of recommendations

Problem : The content-based filtering algorithm using the cosine similarity comparison previously implemented has a limitation in that the accuracy of recommendations is poor and various recommendations cannot be made.

Solution : Using the recbole framework, recommendations based on the neural graph collaborative filtering (NGCF) model provide recipe items that other users like.

Tech Blog

🐍 Using Python in SpringBoot

Problem : It was complicated to operate multiple servers for backend code and ai code when developed in an on-premises environment.

Solution : Integrating back-end and ai servers, and add a 'jython' dependency to use python in spring boot, a back-end framework.

Tech Blog

📜 Requirements Installation

Problem : After I clone the RecBole framework to use in Spring Boot, the torch installation fails while executing the command 'pip install -e. --verbose'.

Solution : The Python version should be between 3.8 and 3.10, the Python and Pip versions must match, and the installed Python should be 64-bit.

Tech Blog

🦾 Preparations for using AI models

Problem : I tried to train the Recbole dataset, but I kept getting this error despite the torch module. 'ModuleNotFoundError: No module named 'torch''.

Solution : Install CUDA and Nvidia Driver, then reinstall PyTorch compatible with the corresponding CUDA version.

Tech Blog