CrowdTI
CrowdTI is a truth inference system in crowdsourcing which makes three contributions. Firstly, since there are dozens of truth inference algorithms, our system has implemented all of these algorithms and can automatically recommend the best inference algorithm to infer the truth of each task. Secondly, our system can help users to infer the worker quality. Thirdly, our system can visualize the results, including the inference quality and time, worker quality, and task statistics.
Dependency (Deployment on OS X)
celery
Follow instructions at here
Briefly you can use brew to install rabbitmq needed by celery.
brew install rabbitmq
After installing, you should alter environment variable PATH as below:
export PATH=$PATH:/usr/local/sbin
Then install celery using pip
pip install celery
If you have problems with celery and django, you can refer to this
gviz_api
pip install -U https://github.com/google/google-visualization-python/zipball/master
mono
Please refer here to install mono, in order to compile and run .NET application.
MATLAB
Please install MATLAB and make sure you can open it in terminal using 'matlab'.
jquery.timers
Have already contained in project. More info here.
Database configuration
The project used django default database, you can use your own as long as you configure it in CrowdInfer/settings.py, you can refer to here for instructions.
Tested Environment
####OS X 10.11
- Python 2.7.10
- Django 1.10.1
- rabbitmq 3.6.1
- celery 4.0.2
- mono 4.2.3
- MATLAB R2014b
Make it Run
Start rabbitmq
Start celery
celery -A CrowdInfer worker -l info
Migrate Django Tables
python manage.py migrate
Then you should load method info from 'main_method.sql'
Run Django
python manage.py runserver
Troubleshooting
If you're encountered with error like
DisabledBackend' object has no attribute '_get_task_meta_for
Make sure you have prepared CrowdInfer/init.py as instructions.