YouTube Comment Sentiment Analyzer

This script intakes a YouTube video link, and outputs all of the video's YouTube comments in an excel sheet, sorted by sentiment (i.e. Positive, Neutral, or Negative). The sentiment analysis done by the 'Hugging Face' cardiffnlp roBERTA-base language model, which was trained on ~58 million tweets.

Table of Contents

• Requirements
• How To Install
• How to Run
• Troubleshooting
• Credits

Requirements

• Libraries
  • Python 3.9
  • googleapiclient 2.14.0
  • openpyxl 3.1.2
  • dotenv 1.0.0
  • transformers 4.35.2
  • tensorflow 2.15.0
  • huggingface-hub 0.19.4
  • langdetect 1.0.9
• Other:
  • Google account
  • YouTube API key

How To Install

• Download the repository from GitHub
  • Click the green 'Code' button
  • Click 'Download ZIP'
  • Unzip the file (find the folder in your computer and right-click 'Extract all')
• Install Python 3.9 to your computer
  • If it does not work, make sure it is added to your system's PATH
  • Note: It must be version 3.9 for Tensorflow to work
• Install Rust
  • Download Visual Studio C++ Build Tools (pre-req for Rust) - https://visualstudio.microsoft.com/downloads/ - Under "All downloads" > "Tools for Visual Studio 2019", find "Build Tools for Visual Studio 2019" and download it. (You can also choose a newer version if available.) - Run the installer - In the installer, select the "C++ build tools" workload - Make sure to include the latest versions of MSVCv142 - VS 2019 C++ x64/x86 build tools and Windows 10 SDK - Restart the computer In the installer, select the "C++ build tools" workload. - https://visualstudio.microsoft.com/visual-cpp-build-tools/
  • https://www.rust-lang.org/learn/get-started
  • If it does not work, make sure it is added to your system's PATH
• Create a Virtual environment
  • https://www.youtube.com/watch?v=APOPm01BVrk&ab_channel=CoreySchafer
  • Open a new terminal and navigate to folder you want to save your virtual env to
  • Make sure pip is installed (to check - run: python -m pip --version)
    • If pip is missing, download pip from the official source: https://bootstrap.pypa.io/get-pip.py
  • Run python -m pip install virtualenv
  • Run python -m virtualenv Comment-Sentiment-Analyzer-Virtual-Env
• Install required libraries in virtual env
  • Open a terminal
  • Activate the virtual environment"
    • Run C:\Users\Michelle\.virtualenvs\Comment-Sentiment-Analyzer-Virtual-Env\Scripts\Activate.ps1
  • Run:
    • pip install --upgrade google-api-python-client
    • pip install openpyxl
    • pip install python-dotenv
    • pip install transformers
    • pip install tensorflow
    • pip install huggingface-hub
    • pip install langdetect
• Open the 'env.txt' file in VSCode (or a text/code editor)
  • Fill in all the 'youtube_api_key' line
    • How to Get a YouTube API Key:
      • Log in to Google Developers Console
      • Create a new project
      • On the new project dashboard, click Explore & Enable APIs
      • In the library, navigate to YouTube Data API v3 under YouTube APIs
      • Enable the API
      • Create a credential
      • A screen will appear with the API key
  • Rename the 'env.txt' file to '.env' and save it
• Update the file paths in 'run_script_1.bat' and 'run_script_2.bat'
  • Note: there are 2 paths per file which need to be updated

How To Run

• Double click the 'run_script_1.bat' file
  • Open the newly created 'YouTube-Comment-Analyzer-Setup.xlsx' file
  • Enter the YouTube video link in the B1 cell
  • Adjust the B2 cell value to analyze more/less comments
    • Note: comment analysis takes ~3 seconds per comment
  • Save and close the file
• Double click the 'run_script_2.bat' file
  • Once it is ready, it will be renamed: "YouTube-Comment-Analyzer-Complete.xlsx"
  • Note: replies to comments will not be recorded
  • Note: only english comments will be recorded
  • Note: the most recent comments will be returned first

Troubleshooting

• The excel files cannot be open while running the script
• Comment analysis takes ~3 seconds per comment, so you may need to wait

Credits

Michelle Flandin