CADOCS: Natual Language Understanding Module

Introduction

CADOCS is a conversational agent working on the Slack platform and able to use third-party tools to identify and manage community smells in software development communities on GitHub.

Specifically, this is the repository containing the Natural Language Understanding service module to interpret users' intent. The main CADOCS app communicates with this module through an API call.

Content of the Repository

The main elements of the repository are described below:

• CADOCS.py: This module is responsible to create the NLU model through the given dataset and to make predictions with it
• cadocs_service.py: This is the web services that exposes the model functionalities through HTTP requests
• dataset.csv: This is the dataset we used to train the model, built with a survey
• default/current: This is the actual model that CADOCS uses to make predictions
• input_testing.ipynb: The Python notebook created to test the model with Input Testing. It contains an execution of the tests
• cadocs_NLU_env.yaml: The Anaconda environment containing the dependencies needed to execute the module

Other Tools

The entire CADOCS tool is composed of three modules:

• CADOCS link: it is the Slack App used to interact with users.
• CADOCS_NLU_Model (this repository): it is the ML service used to interpret the users' intents.
• csDetector link: the augmented and wrapped version of csDetector, used in our tool to detect community smells and other socio-technical metrics.

How to Install CADOCS: Natural Language Understanding Module Locally

Requirements

• Anaconda 3 - 2022.05 - 64bit
• Windows 10
• Microsoft Visual C++ Build Tools 14.0 or higher (In order to run Anaconda)
• (Optional, if you won't use Anaconda to install it) Python 3.7.13

Installation Steps

• Step 1: Local installation of the NLU service (Recommended, using Anaconda)
  • Clone the current repository on your system
  • In our repository, find the cadocs_NLU_env.yaml file which contains the environment and the dependencies needed
  • Through the Anaconda Powershell, run the following command: conda env create -f ENV_FILE_NAME.yaml

In case you faced some error installing the Anaconda environment, please proceed with the Step 1.1

• (Optional) Step 1.1: Local installation of the NLU service (Manual)

  • Clone the current repository on your system
  • In our repository, find the cadocs_NLU_env.yaml file which contains the environment and the dependencies needed
  • Create a new Python 3.7.13 environment
  • Within the previously mentioned file, you will find each of the dependencies needed to run the tool
  • Install by hand each of them in your env

• Step 2: Installation of the English dictionary

  • After having installed the environment, open an Anaconda Powershell with the admin privileges
  • Run the following commands
    • conda activate ENVIRONMENT_NAME
    • python -m spacy download en_core_web_md
    • python -m spacy link en_core_web_md en

• Step 3: Use our Model!

  • Open the project on your system with the IDE you prefer (We suggest using Visual Studio Code or PyCharm)
  • Activate the python environment created in the Step 1 and run the module cadocs_service.py

• (Optional) Step 4: Test it

  • Open your browser
  • Make a GET request like follows: http://localhost:5000/predict?message=YOUR_MESSAGE

Contributors