chisslR

Paper about chissl

YouTube video about chissl

Dustin's chissl powerpoint

Goal

Provide interactive learning during model training of unlabelled data to improve training of the model. See the left red arrow in the below image.

Requirements

Python

conda install flask matplotlib networkx nltk numpy pandas pymongo scipy scikit-learn

conda install -c conda-forge umap-learn

You will need mongodb installed on your system.

brew install mongodb if you are on a Mac or your own way if so preferred

Node
R with the following packages

devtools::install_github("react-R/reactR")

install.packages(c("usethis", "htmlwidgets", "shiny", "shinythemes", "sortable", "cowplot", "textdata", "topicmodels", "tidytext", "dplyr", "tidyverse", "feather", "viridis", "stringr"))

Installation

Once all the above is done go to the chissl/python-chissl folder and run: pip install -e .
Create a mongo db by the name of chissl and then import the downloaded chissl.agz by performing a mongorestore or by using studio3T. This command should work mongorestore --db chissl --gzip --archive=chissl.agz
Start the react app by going to the chissl/react-chissl/ folder and running npm start. Which will start both the flask backend and the react app.
Then run the download_chissl_mongodb command located in helpers/data-prep.R

Standalone

To run the flask backend standalone: Run the backend chissl flask (python) server located in the chissl folder.

python server.py -p 8891 -d -m localhost or try python server.py -p 9101 -d -m localhost

Check if it is working by going to: http://127.0.0.1:8891/api/applications/

Reference code