Unsupervised reviewer-article assignment based on topic models*

Website that implements the reviewer-article assignment system based on Django, Python web framework. You can try an example website here where we run on Amazon EC2 cluster.

*This is work-in-progress. Please use at your own risk

Members

• Daniel Acuna (concept, website, paper writing)
• Titipat Achakulvisut (website, paper writing)
• Tulakan Ruangrong (website)
• Konrad Kording (concept, paper writing)

Requirements

• Django 1.7.4
• Pandas
• Numpy
• Scikit-learn
• gensim
• unidecode
• Django Crispy Forms
• Django Tables 2
• Glop (Google's linear programming solver)
• Celery
• Django-celery
• rpy2
• R with the ARM package
• mpld3

Installation

We provide requirements.txt for all the requirements that we need. After cloning repository, you can do:

pip install -r requirements.txt

In addition of the requirement, we need to install or-tools which has installation document on this link. For Mac OSX, you might find problem installing or-tools where we solve that using easy-install (see this issue)

For the article scoring system, you need to install the R programming language and the arm library using the command

install.packages('arm')

Usage

Local usage to launch the website can be created by hovering to where manage.py located. Then, you can launch celery worker by running:

celery -A reviewer_assignment_website worker

Afterward, you can run the server as follows:

python manage.py runserver

License

(c) 2014-2015 Daniel Acuna, Titipat Achakulvisut, Tulakan Ruangrong, and Konrad Kording