search

ufal / factgenie

Data visualization and span-based annotation of generated texts
MIT License
15 stars 1 forks source link

Support other crowdsourcing services #18

Open kasnerz opened 3 months ago

kasnerz commented 3 months ago

Currently, our annotation framework targets Prolific although we are not affiliated with them in any way.

We should make it clear that we support any crowdsourcing service that follows the similar workflow.

This means updating the README and changing the hard-coded Prolific stuff (mostly the final page in the default annotation template, see annotate_default.html).

kasnerz commented 1 week ago

This is partially solved now. The final message can be customized for each crowdsourcing campaign, including the link shown to the annotators. The instructions on wiki are also now quite general.

What still needs to be investigated are the URL parameters. In factgenie/main.py:annotate, we currently expect the parameters PROLIFIC_PID, SESSION_ID and STUDY_ID that are specific to Prolific.

To properly support other crowdsourcing services, we should enable customizing these parameters.