We have a wiki for this project at http://dev.stanford.edu/confluence/display/collaborators/DeepDive+Genomics

Contact Matteo Riondato rionda@cs.stanford.edu for information.

We use a GUI tool called Mindtagger to expedite the labeling tasks necessary for evaluating the data products of DeepDive.

To start the GUI tool, simply run:

./labeling/start-gui.sh

and open a browser to "http://localhost:8000".

The GUI shows, in the upper left corner, all tasks defined under labeling/ directory. The results of the labeling (including intermediate data) are stored under each task directory.

Use the push.sh script:

./push.sh

It will create a commit holding any changes made on your side, and push to the original GitHub repo. Enter your GitHub user name and password that has been added as a collaborator account when asked. If you don't have a collaborator GitHub account, the script can generate a set of patch files that can be emailed to the authors instead.

We will review and merge the work, and it will help the next iteration of improving the data products.

Use the pull.sh script to grab any new tasks or changes made on the original repo:

./pull.sh

It will first create a commit holding any changes made on your side, and fetch and merge the updates, honoring the changes you made (read: the tags you entered).

Use the labeling/create-new-task.sh script to create new labeling tasks.