Discussion: Packaging phenomena-annotated corpora (Moderator: Ned, Scribe: Prescott) DELPH-IN Summit 2013 7/30/13


Ned: Biggest roadblock to packaging is character info

Woodley: single or discontinuous spans?

Ned: could have discussion about that, may need to consider both.

Stephan: string value field will be added to tsdb profiles

Ned: any other metadata needed? template for resources

Dan: remind us how you are thinking about choice of vocabulary. Where does synonymy live in universe of phenomena annotation?

Ned: haven’t prescribed how to document. More interested in phenomena detection. Standard like GOLD for interoperability. A bit too high level/coarse grained. Would have been nice.

Dan: Put a field in annotation scheme for canonical reference for phenomena.

Ned: could be a difficult thing, but may also be good for interoperability

Emily (to Dan): do you mean a slot in profile?

Dan: don’t know

Ned: good suggestion.

Sanghoun: are you thinking about web interface for viewing phenomena inventory and visualization?

Ned: no haven’t thought about that. Ned using BRAT for visualization

Sanghoun: worried about format for web viewing/linking would like to know what direction Ned will go

Ned now moves beyond discussion to talk about graphs demonstrating phenomena work. He is showing slides of the graphs.

Rebecca: one way of describing a domain

Ned: could be a good way of characterizing domain of phenomena

Tim: Looking not a phenomena, but what are the differences

Last update: 2013-07-31 by PrescottKlassen [edit]