Transliterate into English about polyglot HOT 3 CLOSED

Please check the version of the software available on github.

On Thu, Jun 9, 2016, 12:01 AM Amir Othman [email protected] wrote:

Hi,

firstly, thanks for this great software. I realized that in the code, that
when the target language is English, it will not transliterate it and
returns the original text. Does polyglot does not support transliterating
into English or am I missing something?

regards,
Amir

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#57, or mute the thread
https://github.com/notifications/unsubscribe/AAKgliju0HyY2kjcNMs7f710lyJirpmzks5qJ7pcgaJpZM4Ixru6
.

from polyglot.

Hi,

I did use the software available from github. I installed the new version with the following command:

pip install -U git+https://github.com/aboSamoor/polyglot@master

Here is a gist demonstrating what is happening.

But this behaviour is somehow expected because in the code transliteration/base.py, the decoder would be set to '_dummy_coder', if the target language is 'en', which would return the original text as is.

Which I find a bit unclear to me why is it doing so. Any form of help would be greatly appreciated.

Thanks again for the reply.

regards,
amir

from polyglot.

Hi, @amirothman

Regarding transliteration

Polyglot transliterates when the target languages is English. For example, when source language is Russian, it works like:

>>> Text("Язы́к — сложная знаковая система, естественно или искусственно созданная и соотносящая понятийное содержание и типовое звучание (написание).").transliterate(target_language="en")

WordList(['', '', 'slojnaya', 'znakovaya', 'sistema', '', 'estestwenno', 'ili', 'iskusstwenno', 'sozdannaya', 'i', 'sootnoxiaszczaya', 'poniatiinoe', 'soderjanie', 'i', 'tipovoe', 'zvuchanie', '', 'napisanie', '', ''])

Regarding your example

The model for transliterating Malay is learnt to producing identical transliteration. The fact that a language is written in Latin alphabet doesn't mean it will change through transliteration.

Regarding your idea on decoder

In transliteration code we have a encoder and a decoder, which means we transliteration from source language to English using the encoder, and then from English to the target language using the decoder. So it is natural to set encoder to _dummy_coder if the source language is already English, and to set the decoder to _dummy_coder if the target language is English.

from polyglot.