Zoom chat analytics

Author

Lindsey Viann Parkinson

Purpose

When a Zoom meeting is recorded both the video and a transcript of the chat box is included. The transcript is output in the form of a .txt file. The Zoom_chat_analytics repository is the beginning of a tool to harness more information from those meeting transcripts. This includes assessing speaker engagement, participant involvement over time, participant clusters, and in the future sentiment.

All the analytical functions can be run from the command line.

As of December 17, 2020 Zoom_chat_analytics.py can:

• Create and export a csv of the chat with: time of comment, author, and comment
• Create and export a pdf that includes the following:
  • Ordered list of which participants commented the most
  • A histogram of comment volume of the course of the meeting
  • Any URLs shared
  • A network graph of commenters and responders
• Each of the analytical functions within the pdf report can also be run individually

Documents

• requirements.txt - installs all required dependencies
• Zoom_chat_analytics.py - see above

Example output

Examples of the histogram, network graph, and basic pdf report are in the examples folder

How to use this Repo

The command line application runs in the terminal using Python Fire. In order to use the application you will need to install the required dependencies listed in the requirements file.

pip3 install -r requirements.txt

If you only want to make the csv:

• convert_to_csv(filepath, csv_name) - Creates and exports three column csv: time, author, comment

If you want to create the csv and run analysis:

• start_to_finish(filepath, csv_name, pdf_name) - Runs all the analytical functions and exports pdf, dataframe csv, and graphs as .png files
• one_analytic(filepath, csv_name, function_name) - Runs one analytical function
  • find_urls - Extract urls from comments
  • comments_by_author - Count the number of comments made by each author
  • comments_over_time - Returns histogram of comments over meeting time
  • comment_network - Create a network graph of who responds to comments

Variables:

• filepath: path to .txt file
• csv_name: name for exported csv (include .csv)
• pdf_name: name for exported pdf (include .pdf)
• function_name: which of the individual functions you'd like to run

In the terminal:

• Change directory to where Zoom_chat_analytics.py is stored
• The three parts of the application call are $ {./Zoom_chat_analytics.py} {function name} {function variables}
  For example:

./Zoom_chat_analytics.py start_to_finish /home/folder/folder/Zoom.txt example.csv example.pdf

Further reading

My goal is to create a tool that allows someone to glean more insights from their meetings. More about my goals for the application can be found in this Medium post If you are interested in contributing check out the What's next section below for a list of big and little projects related to the application. Or go crazy on your own idea!

What's next

New Function(s)

• Sentiment analysis by author
• Sentiment analysis over time

Improvements, bugs, tweaks, expansions, etc.

• find_urls When it runs as a single function, make it optional to send to pdf
  • currently prints links to terminal
  • can create a small create_pdf function to call? Or a wrapper for multiple functions?
• comments_over_time Convert matplotlib histogram to seaborn histogram so it's prettier
• create_csv Add option for keeping private messages
• create_csv change regex call to ignore "_:" with author names. Then remove code that makes the change later in the function
• comments_by_author Make the number of comments line up in the output
• comments_by_author When it runs as a single function, make it optional to send to pdf
• start_to_finish Find better pdf creator? Make the pdf ouput more professional.
• start_to_finish create option to export only certain functions

Citation

If you use this repo please consider citing it.

@misc{Zoom_chat_analysis,
  title="A terminal application for analytics of Zoom chat transcripts",
  author="Lindsey Viann Parkinson",
  year="2020",
  publisher="Github",
  journal="GitHub repository",
  howpublished= "https://github.com/LVParkinson/Zoom_chat_analytics"
}