search

thiborose / gecko-app

A web application that interfaces two GEC systems. [web instance is down]
https://gecko-app.azurewebsites.net/
Other
31 stars 9 forks source link

Detokenization doesn't work as expected #10

Closed jacqle closed 3 years ago

jacqle commented 3 years ago

Current behavior: the TreebankWordDetokenizer fails to group back together hyphenated words. Expected behavior: the output text does not contain any additional spaces compared to the input text. E.g.: An out-of-the-box feature -> An out - of - the - box feature

jacqle commented 3 years ago

One possible fix would be to add an ad hoc regex.

Dodo-s95 commented 3 years ago

One possible fix would be to add an ad hoc regex.

yes, that could work. I think it's the quickest thing we can do.