NIS-2018-CROSS-M / cross-lingual-morph-analysis

Create morphosyntactic annotations for low-resources languages
GNU General Public License v3.0
0 stars 1 forks source link

Train models and think how to load them on a server for frontend creation #30

Open erba994 opened 4 years ago

erba994 commented 4 years ago

think about how to organize the frontend. The idea is to include frequency files with annotations for the 3 language families as an input for the model, a list of words and the indication of the language from the user and the tagged words as an output (possibility to choose Apertium or UDPipe format?)