agenda for meeting (6PM CEST) on unimorph / shared task on inflection

[Ryan Cotterell]

Hello,

One important update: We have now created a UniMorph steering committee. The members are currently: Ekaterina Vylomova, Ryan Cotterell, Mans Hulden, and David Yarowsky. The first meeting was last Friday, and the group will meet once a month to start off. You can email the committee at unimorph...@googlegroups.com. We will work on developing some sort of protocol for expanding the steering committee and making transparent decisions. (This move was inspired by the Universal Dependencies (UD) project.)

Meeting Today:

• Zoom link (same as Google calendar invite):https://ethz.zoom.us/j/96798828450
• Collective Google doc: https://docs.google.com/document/d/10Y-lNfyzRMOhgqOnVEK5Ry9SIzTEcaxL80A8S-dUXNI/edit?usp=sharing. The Google doc will contain the minutes of today's meeting as well as thoughts and suggestions expanded after the meeting. The idea of having a public document is to make attendance in any one meeting not necessary and to be able to have multiple meetings to fit everybody in at different times without losing the common thread. One could contribute to the Google doc without attending the Zoom call at all. 

Agenda for Today's Meeting (and the one on Friday):

1. How to modularize the UniMorph annotation effort? The goal here is to decentralize our annotation effort so that we have focused working groups on individual language families. This structure mirrors the UD organizational structure; indeed, the only difference is that UniMorph intends to have working groups at the language family--level rather than at the language level. These focused working groups would function autonomously and, hopefully, people will organically take ownership of producing high-quality, publicly available morphological resources.
2. UniMorph's citration and publication strategy: We need to make sure people who use our published resources cite the original work, and not just our paper in many cases. The UniMorph 3.0 paper at LREC 2020 got into some hot water over this issue, so it would be best to avoid the problem in the future. I think it also makes sense to move away from a single UniMorph paper with an ever-growing list of authors. We should, instead, publish a few papers with far fewer authors that deal with language-specific issues, e.g. a paper published by those who did the conversion of languages in a given family, e.g. Nilotic, and maybe a larger paper with overarching changes to the schema. This will help people get credit for what they did. I am not sure a 100-author paper helps anyone.
3. Shared task: The goal here is to determine how the shared task we will host this year can and interact with and encourage this year's UniMorph annotation effort. Please also respond if you would like to join our shared task Slack where more focused discussion is taking place.

Cheers,

--Ryan