Issue-Assigner(Chinese name: Lingpei)

Issue-Assigner is a comprehensive solution designed to automate the assignment of issues in a repository. It integrates seamlessly with your development workflow to recommend the most suitable developers for resolving specific issues, thereby enhancing efficiency and collaboration.

Issue-Assigner Framework

Features

1. Data Acquisition and Updates

• Real-time Data Acquisition: Utilize the GitHub REST API to obtain the latest collaboration data within the repository.
• Continuous Updating: Support incremental training of the model to adapt to dynamic changes in data, meeting the current application needs of repository maintainers.(feature under development)

2. Issue Assignment Task Benchmark

• Unified Framework: Implements various models for training and testing under a unified framework using PyTorch Geometric (PyG).
• Module composition: Includes dataset construction, data loading, and model design tailored for issue assignment tasks.

3. Frontend Plugin Functionality

• Display Recommendations: Shows recommended developers for the current issue directly within the issue interface.
• Model Selection: Allows users to choose from different available models to view various recommendation results.
• Metric Display: Show the Activity of developers over the past three months (weighted exponentially with more recent time having greater weight), average Community_OpenRank contribution, and average Global_OpenRank influence. This helps repository maintainers make more informed decisions. Refer to the Opendigger's doc for the meanings of these metrics.
• Feedback Mechanism: Users can provide feedback on the recommendations using a thumbs-up (approve) or thumbs-down (disapprove). (This project will not be realized. For details, please refer to Issue-Assigner-BaaS.)

4. Backend Service Functionality

• Result Storage: Stores the recommended results post-model testing for frontend display.
• Feedback Storage and Evaluation: Records real user feedback from the frontend and utilizes the OpenRank for subsequent model evaluation (feature under development). (This project will not be realized. For details, please refer to Issue-Assigner-BaaS.)

Project Structure

├── config/             # Model configuration files
├── data/               # Data loading
├── dataset/            # Dataset construction and preprocessing
├── frontend/           # Frontend plugin files
├── model/              # Model implementation, training, and testing
├── server/             # Backend service using FastAPI
├── tools/              # Some toolkits, such as NLP processing and logging tools
├── IssueAssign.py      # IssueAssign class for unified management of models
├── main.py             # Entry point for model training and testing
├── LICENSE             # Open source software license
├── requirements.txt    # Python dependencies
└── README.md           # Project documentation

Installation

Prerequisites

• Python: 3.10.15
• Torch: 1.13.1
• Git: To clone the repository
• Node.js and npm: For frontend development (if you plan to modify the frontend)
• Browser: Currently only tested on Edge
• MongoDB: 7.0.1
• Neo4j: 5.12.0

Clone the Repository

git clone https://github.com/zhingoll/Issue-Assigner.git

Install Dependencies

pip install -r requirements.txt

Usage

Getting the Specified Repo Data

First, you need to create a config.yaml file in the data folder. Examples are as follows:

mongodb:
  db: "your_db_name"
  url: "your_db_url"
  tz_aware: true
  uuidRepresentation: "standard"

tokens:
  - "your_token1"
  - "your_token2"
  ...

projects:
  - "owner/name" # The repo name you want to get, such as 'X-lab2017/opendigger'
  ...

neo4j:
  uri: "your_uri"
  username: "your_neo4j_name"
  password: "your_password"

repo: # example
  owner: "X-lab2017" 
  name: "open-digger"  # The name(owner/name) of the repo you want to make into a dataset

After creating the config.yaml file, run:

python .\data\repo_data_collect.py 

This command will get the data of your specified repo and store them in mongodb.

Building 'Developer-PR-Issue' Collaboration Graph

Run the following command to import the data stored in mongodb in the previous step into neo4j and build a 'Developer-PR-Issue' collaboration graph.

python .\data\mongo2neo.py

Generating the Raw Materials to Build the Dataset

Run the following command to generate the raw materials needed for model training datasets.

python .\data\issue_content.py
python .\data\neo_mongo2csv.py opendigger

Once the script has run, it will automatically create a new folder named after the specified repository within the dataset directory. Additionally, within this new folder, the script will also create a subfolder named raw. All generated CSV files will be automatically saved in this raw folder, thus being immediately ready for use in subsequent model training dataset construction. Here are specific examples of where the CSV files will be saved. If the repository name specified is "opendigger," then the paths for the saved CSV files will appear as follows:

dataset\opendigger\raw\user_issue.csv 
dataset\opendigger\raw\open_issues.csv 
dataset\opendigger\raw\resolved_issues.csv 
dataset\opendigger\raw\issue_content.csv

Running the Main Application

Navigate to the project's root directory and run:

python main.py

This command initiates the training and testing of models related to the issue assignment tasks.

Obtain Metric Data

Run the following command to retrieve the developer Activity in a specified repository over the past three months, as well as the Community_OpenRank contribution and the Global_OpenRank value. The Activity is calculated using an exponentially weighted sum, giving greater weight to more recent activity. Other metrics are calculated as averages.

python .\server\developer_metrics.py

Using the Frontend Plugin

1.Load the Plugin in the Browser

• Open the Edge browser (currently tested only on Edge).
• Enter developer mode by navigating to edge://extensions/.
• Enable the Developer mode toggle.
• Click on Load unpacked and select the frontend folder from the project directory.

2.Start the Backend Service

In a new terminal window, navigate to the server directory:

cd server

Start the FastAPI service using Uvicorn:

uvicorn server:app --reload

3.Interact with the Plugin

As the project is still in the experimental stage, the suggested issue is specified in the file opened_issues.csv. The project path example for this file is: dataset\opendigger\raw\. You can experience the functionality of the plugin by using the issue number provided in this file.

• Use the provided interface to select different models and view their recommendations.
• Provide feedback using the thumbs-up or thumbs-down icons. (This project will not be realized. For details, please refer to Issue-Assigner-BaaS.)

Future Plans

• Model Expansion: Incorporate additional models and algorithms to improve recommendation accuracy.
• Browser Compatibility: Extend plugin support to other browsers like Chrome and Firefox.
• Simplify the project use process and make it more convenient for users.
• Feedback Integration: Utilize user feedback combined with the OpenRank algorithm for enhanced model evaluation (feature under development).

Contact

If you have any questions or suggestions, feel free to open an issue~

License

This project is licensed under the Apache-2.0 license, please make sure abide by the licenses when using the project.