yarakyrychenko / west-conversations

A template repository for linguistic analysis of conversations from scratch using Convokit.

west-conversations

⭐ A template repository for linguistic analysis of conversations from scratch using ConvoKit.

Goal

This repository provides a quick way to analyze custom conversational datasets by first converting them into a ConvoKit Corpus object. The ConvoKit module has many interesting linguistic analyses you can run on your conversation data. This template covers creating a corpus from your data (MakeConvokitCorpus.ipynb), basic descriptive statistics (DescriptiveStatistics.ipynb), linguistic coordination (LinguisticCoordination.ipynb), and politness strategies speakers employ (Politeness.ipynb).

Using the Template

• Step 1: Click the green button that says Use this template at the top.
• Step 2: Follow the instructions on the screen to create your repository.
• Step 3: Clone it to your local machine (you can use GitHub Desktop).
• Step 4: Go to .gitignore and uncomment data/ to keep your data private at all times.
• Step 5: Upload your data (master.csv and transcritps) to your local repostiory. There is an example master.csv in this template to help you with formatting.
• Step 6: Run MakeConvokitCorpus.ipynb to create your corpus.
• Step 7: Check out the python notebooks in the analysis folder.

Template Structure

.
├── data
│   ├── processed
│   │   └── corpus
│   └── raw
│       ├── master.csv
│       └── transcripts
├── processing
│   └── MakeConvokitCorpus.ipynb
├── analysis
│   ├── LinguisticCoordination.ipynb
│   ├── DescriptiveStatistics.ipynb
│   ├── Politeness.ipynb
│   └── utils.py
├── results
├── viz
└── README.md