Clockwork Raven

Human-Powered Data Analysis

Clockwork Raven is a web application that allows users to easily interact with Mechanical Turk. It is actively used at Twitter to gather thousands of judgments every day.

If you've ever wanted a friendlier interface to Mechanical Turk, then Clockwork Raven is for you.

What it is good for

Clockwork Raven is designed for individuals, or organizations that share a single Mechanical Turk account. Users can upload data, design evaluations to send to Mechanical Turk through a simple, drag-and-drop interface, and review results all in a single place. It also provides a Thrift API for users who wish to programmatically run human computation tasks or gather real-time judgments.

To control the quality of responses, Clockwork Raven allows you to restrict tasks to Mechanical Turk's "Categorization Masters", or to only those users whom you have marked as Trusted in Clockwork Raven.

Setup

Check out the code from https://github.com/twitter/clockworkraven
Requirements:
1. Make sure the machine that you're using has Ruby 1.9.3 installed. The easiest way to install and manage Ruby is with RVM.
2. You'll need the RubyGem "bundler" installed, and then just run bundle install from the Clockwork Raven directory to install all of the libraries needed by Clockwork Raven.
3. Clockwork Raven uses Resque to run tasks in the background. Resque requires a Redis server -- see Resque's instructions for installing Redis. By default Clockwork Raven assumes your Redis server is running on localhost:6379. If this isn't the case, edit config/resque.yml.
4. In a production environment (e.g. any environment where Clockwork Raven will be accessible to users), it should be run over SSL to protect users' credentials when they log in. If you don't use SSL, these credentials will be sent over the network in the clear!
Configure:
1. Generate a secret key. Copy config/secret.example.yml to config/secret.yml. Then, run rake secret and copy the output to config/secret.yml.
2. Copy config/database.example.yml to config/database.yml and modify it to point to your MySQL database. Currently, Clockwork Raven only supports MySQL.
3. Copy config/mturk.example.yml to config/mturk.yml. Follow the instructions in that file to connect Clockwork Raven to your Mechanical Turk account.
4. Configure authentication:
  
  LDAP Authentication
  
  LDAP authentication is the recommended way to manage account in Clockwork Raven. If your LDAP server supports SSL/TLS, copy config/auth.example_ldap_encrypted.yml to config/auth.yml. If your LDAP server does not, copy config/auth.example_ldap_unencrypted.yml to config/auth.yml. Follow the instructions in that file to connect Clockwork Raven to your LDAP server and grant access to specific LDAP groups and users.
  
  Password Authentication
  
  If you can't use an LDAP server, you can configure Clockwork Raven to use "password authentication," which will allow you to manually create accounts. Copy config/auth.example_password.yml to config/auth.yml. Then, you can create accounts by running "rake users:add" and change passwords with rake users:change_password. Note that you will need to set up your database (explained below) before using these rake tasks.
Set up the database. If the databases you configured Clockwork Raven to use in config/database.yml do not exist, run rake db:create to create them. Then, run rake db:structure:load to load the database structure into your database.
Start up the background workers. Just run rake raven:resque to start up 4 background workers. You can start up more background workers by passing an argument to the rake task: rake raven:resque[16] will start up 16 background workers.
Run the server. To run the server, run rails server.

Security

Administrators can separate users into "privileged" and "unprivileged" sets manually or based on LDAP groups. Unprivileged uses are not allowed to spend money, but can submit evaluations to the Mechanical Turk sandbox to test out the system.

Clockwork Raven is designed for situations where everyone who has access to the system is relatively trusted. Because the form designer allows users to use arbitrary HTML, anyone with access to the system could execute an XSS attack and compromise the system.

Documentation

Documentation is available on the wiki.

Contact

Follow @clockworkraven for updates and notifications. Submit bug report and feature requests to the issue tracker.

Join the mailing list, twitter-clockworkraven@googlegroups.com, on Google Groups to ask questions and discuss development.

Roadmap

We would love any help adding ideas or implementing them!

JSON/REST API.
Provide the option to have multiple Mechanical Turk users complete each task.
Provide in-depth analytics about workers and automate the process of choosing trusted workers.

Contributing

To contribute to Clockwork Raven, fork the repo, make your changes, and submit a pull request. All pull requests should be against *-wip branches. Nothing gets committed/merged directly to master. To merge your pull request, you'll need to include appropriate documentation and tests. Get in touch if you have any questions about what you need to do to get your contributions accepted.

Authors

Ben Weissmann, @benweissmann
Edwin Chen, @echen
Dave Buchfuhrer, @daveFNbuck

Versioning

The current version is in the VERSION file and accessible in the code as ClockworkRaven::VERSION. Releases will be tagged with their release number in Git.

Clockwork Raven uses semantic versioning. Basically, this means that versions will be of the form X.Y.Z, where X is the major version (incremented when backwards-incompatible changes are introduced), Y is the minor version (incremented when backwards-compatible features are introduced), and X is the patch number (incremented when backwards-compatible bug fixes are introduced). Note, however, that these are only hard rules once Clockwork Raven reaches 1.x. Until then, we will do our best to adhere to these policies (particularly with regards to not introducing backwards-incompatible changes in patch releases), but we may make backwards-incompatible changes while only incrementing the minor version number.

License

Licensed under the Apache License, Version 2.0: http://www.apache.org/licenses/LICENSE-2.0

simonasta / clockworkraven