YerevaNN / translit-rnn

Automatic transliteration with LSTM
92 stars 20 forks source link

statistics in transliteration.json? #14

Open Moadab-AI opened 5 years ago

Moadab-AI commented 5 years ago

Just wondering how do you come up with romanization rules statistics, i.e the probabilities p(Latin char | Armenian char) ? of course without having to go through a significant amount of "real" romanized text, labeling them and counting.

Hrant-Khachatrian commented 5 years ago

Dear @moabaom ,

We compiled this list manually using our intuition, there is no scientific basis for these numbers.