Gargantext with Purescript (FrontEnd instance)
About the project
GarganText is a collaborative web-decentralized-based macro-service platform for the exploration of unstructured texts. It combines tools from natural language processing, text-data-mining tricks, complex networks analysis algorithms and interactive data visualization tools to pave the way toward new kinds of interactions with your digital corpora.
This software is free software, developed and offered by the CNRS Complex Systems Institute of Paris Île-de-France (ISC-PIF) and its partners.
GarganText Project: this repo builds the frontend for the backend server built by backend.
Getting set up
There are two approaches to working with the build:
- Use our Nix or Docker setup
- Install our dependencies yourself
With Nix setup
First install nix:
sh < (curl -L https://nixos.org/nix/install) --daemon
Verify the installation is complete
$ nix-env
nix-env (Nix) 2.3.12
To build the frontend just do:
nix-shell --run build
Just serve dist/index.html with any server and you are ready to be connected to any backend.
With Docker setup
You will need docker and docker-compose installed.
First, Source our environment file:
source ./env.sh
WARNING: you must source ./env.sh
before using the docker
container. If you don't do that, the container will write files as
root and you'll need root powers to get ownership back!
Now build the docker image:
docker-compose build frontend
That's it, skip ahead to "Development".
Manual setup
The build requires the following system dependencies preinstalled:
- NodeJS (11+)
- Yarn (Recent)
NodeJS
On debian testing, debian unstable or ubuntu:
sudo apt update && sudo apt install nodejs yarn
On debian stable:
curl -sL https://deb.nodesource.com/setup_11.x | sudo bash -
sudo apt update && sudo apt install nodejs
On Mac OS X with homebrew:
brew install node
For other platforms, please refer to the nodejs website.
Yarn (javascript package manager)
On debian or ubuntu:
curl -sS https://dl.yarnpkg.com/debian/pubkey.gpg | sudo apt-key add -
echo "deb https://dl.yarnpkg.com/debian/ stable main" | sudo tee /etc/apt/sources.list.d/yarn.list
sudo apt update && sudo apt install yarn
On Mac OS X with homebrew:
brew install yarn
For other platforms, please refer to the yarn website.
Development
Docker environment
Are you using the docker setup? Run this:
source ./env.sh
This enables the docker container to run as the current user so any
files it writes will be readable by you. It also creates a darn
shell alias (short for docker yarn
) for running yarn commands inside
the docker container.
Basic tasks
Now we must install our javascript and purescript dependencies:
Note: if you're installing manually you might also need to manually install psc-package
darn install -D && darn install-ps # for docker setup
yarn install -D && yarn install-ps # for manual setup
You will likely want to check your work in a browser. We provide a local development webserver that serves on port 5000 for this purpose:
darn server # for docker setup
yarn server # for manual setup
To generate a new browser bundle to test:
darn build # for docker setup
yarn build # for manual setup
If you are rapidly iterating and just want to type check your code:
darn compile # for docker setup
yarn compile # for manual setup
You may access a purescript repl if you want to explore:
darn repl # for docker setup
yarn repl # for manual setup
If you need to reinstall dependencies such as after a git pull or branch switch:
darn install -D && darn install-ps # for docker setup
yarn install -D && yarn install-ps # for manual setup
If something goes wrong building after a deps update, you may clean build artifacts and try again:
# for docker setup
darn clean-js # clean javascript, very useful
darn clean-ps # clean purescript, should never be required, possible purescript bug
darn clean # clean both purescript and javascript
# for manual setup
yarn clean-js
yarn clean-ps
yarn clean
If you edit the SASS, you'll need to rebuild the CSS:
darn css # for docker setup
yarn css # for manual setup
A guide to getting set up with the IDE integration is coming soon.
Testing
To run unit tests, just run:
test-ps
Note to contributors
Please follow CONTRIBUTING.md
How do I?
Add a javascript dependency?
Add it to package.json
, under dependencies
if it is needed at
runtime or devDependencies
if it is not.
Add a purescript dependency?
Add it to spago.dhall
(or run spago install ...
).
If is not in the package set, you will need to read the next section.
Add a custom or override package to the local package set?
You need to add an entry to the relevant map in
packages.dhall
. There are comments in the file explaining how it
works. It's written in dhall, so you can use comments and such.
Theory Introduction
Making sense of out text isn't actually that hard, but it does require a little background knowledge to understand.
N-grams
N-grams in contexts (of texts) are at the heart of how Gargantext makes sense out of text.
There are two common meanings in the literature for n-gram:
- a sequence of
n
characters - a sequence of
n
words
Gargantext is focused on words. Here are some example word n-grams usually extracted by our Natural Language Process toolkit;
-
coffee
(unigram or 1-gram) -
black coffee
(bigram or 2-gram) -
hot black coffee
(trigram or 3-gram) -
arabica hot black coffee
(4-gram)
N-grams are matched case insensitively and across whole words removing the linked syntax if exists. Examples:
Text | N-gram | Matches |
---|---|---|
Coffee cup |
coffee |
YES |
Coffee cup |
off |
NO, not a whole word |
Coffee cup |
coffee cup |
YES |
You may read more about n-grams on wikipedia.
Gargantext allows you to define and refine n-grams interactively in your browser and explore the relationships they uncover across a corpus of text.
Various metrics can be applied to n-grams, the most common of which is the number of times an n-gram appears in a document (occurrences). GarganText uses extensively the cooccurrences: times 2 n-grams appear in same context of text.
Glossary
document
: One or more texts comprising a single logical document
field
: A portion of a document or metadata, e.g. title
, abstract
, body
corpus
: A collection of documents as set (with no repetition)
n-gram/ngram
: A word or words to be indexed, consisting of n
words.
This technically includes skip-grams, but in the general case
the words will be contiguous.
unigram/1-gram
: A one-word n-gram, e.g. cow
, coffee
bigram/2-gram
: A two-word n-gram, e.g. coffee cup
trigram/3-gram
: A three-word n-gram, e.g. coffee cup holder
skip-gram
: An n-gram where the words are not all adjacent. Group 2 different
n-grams to enable such feature.
k-skip-n-gram
: An n-gram where the words are at most distance k from each other. This
feature is used for advanced research in text (not yet supported in
GarganText)