uhh-lt / context-eval Goto Github PK

Hello, I would like to thank you for your great work.

I need to reference this work in my manuscript paper, I added in the footnote the link to this repository Git Hub, however, the reviewer asked to give a proper citation. Can you help me with how to do that, please?

Best regards,
Asma

I've noticed that the sense mapping between a custom inventory and TWSI inventory is saved to a file with unique name bound to file name of custom inventory.

The script could check for existence of such a file and load mapping instead of computing them anew. It would be a great improvement for multiple evaluations with the same inventory, cause mapping often takes more time then predictions evaluation itself.

Possible drawbacks are:

1. Mapping files pile up, get forgotten and take too much space
2. By chance two different custom inventories with the same name are used at different times -> old mapping taken instead of a new one
   Alternative solution:
   optional parameter for a path to a mapping file?

I would be happy to learn your opinion about it.

data/Dataset-TWSI-2-GOLD.csv.gz has more examples than data/Dataset-TWSI-2.csv which should be processed by WSD systems. This probably means that it is impossible to get 100% coverage.

this URL is not exist https://www.lt.informatik.tu-darmstadt.de/de/data/twsi-turk-bootstrap-word-sense-inventory/