Full documentation is available here
The SWAP algorithm was developed by Phil Marshall et al., and the original implementation of the algorithm that provided the motivation for this work can be found here.
The caesar_external library provides functionality for interacting with live classifications from Panoptes and Caesar in the Zooniverse. Installing this library activates these features in swap.
swap new ${NAME} where ${NAME} is arbitrary. To configure the annotation
parser, pass the --config flag, and then enter the desired configuration.
For example, for Supernova Hunters, enter:
config.annotation.update({'task': 'T1', 'true': ['Real', 'Yes', 1], 'false': ['Bogus', 'No', 0]})
Start by initializing swap's gold labels by running swap golds ${NAME} ${GOLDS_CSV_FILE} where ${NAME} is the name used to configure swap, and the
golds csv file contains the gold labels with column labels 'subject', and
'gold'.
Then run swap on a panoptes csv dump with swap run ${NAME} ${CLASSIFICATION_DUMP}.
If memory is an issue, split the csv dump into batches, and run
swap.truncate() in the python terminal between batches. This truncates the
score history from the user and subject agents, but maintains enough
information for the swap algorithm to continue functioning.
Online swap queries classifications from the panoptes api and sends them
to a placeholder reducer in caesar. Make sure caesar_external is properly
setup for this project, then configure online swap with:
swap online config ${NAME} ${CAESAR_NAME} Where ${CAESAR_NAME} is the name
used to configure the caesar_external library for this project.
Then run swap online with swap online run ${NAME}.