The Python API has two methods for segmentation: Adding morphemes and adding morph boundaries / forbidding morph boundaries (adding known-morpheme-internal places).
But when serializing the data to the DeriNet 2.0 file format, only morphemes get saved and the boundaries are lost.
It might be possible to save at least the positive boundaries as Implicit morphemes. Another possibility would be to save them as morphemes with only one of Start / End filled.
Saving the forbidden boundaries is difficult and might need new syntax support.
The Python API has two methods for segmentation: Adding morphemes and adding morph boundaries / forbidding morph boundaries (adding known-morpheme-internal places).
But when serializing the data to the DeriNet 2.0 file format, only morphemes get saved and the boundaries are lost.
It might be possible to save at least the positive boundaries as
Implicitmorphemes. Another possibility would be to save them as morphemes with only one ofStart/Endfilled.Saving the forbidden boundaries is difficult and might need new syntax support.