iink SDK on Device

Answered

Supporting 2 languages with 2 editors

This is a follow up to the following thread:

https://developer-support.myscript.com/support/discussions/topics/16000029686/page/last#reply-to-post

The issue is that when configuring an editor to use Japanese as its language, the editor can recognize text written in Japanese and in English, but the English recognition is poor compared to when using en_US as the editor's language.

We would like a way to be able to recognize Japanese text, but to allow the user to force recognition using the en_US configuration when the user is not happy with the recognition result of English words.

We discussed transferring data between 2 editors: one in English (not visible by the user) and one in Japanese (visible and used by the user). I actually got a chance to try to implement this, but I'm stuck

The suggested solution was to export and import the data as MimeType.TEXT between the editors, but this requires obtaining the text after doing text recognition and conversion. So I would basically be importing text that was wrongly recognized by the Japanese editor to the English editor, which does not help with improving the recognition.

What I think I need is a way to export the user's strokes, before recognition, from the Japanese editor, then import that data into the English editor, do recognition there to get English text, and finally import that text data back into the Japanese editor to display it to the user.

Is this something that is possible? I was thinking that maybe exporting and importing as JIIX might work but the documentation about "Import and export formats" mentions this for importing JIIX to a Text block: "For the time being, JIIX import of text content can only be used to select recognized text word candidates". So I'm not sure it can be used for our purposes.

Is there some other, maybe better way to achieve the original goal?


Best Answer

Dear Nicolas,


thank you for the precisons.


To answer your questions, indeed, the only way JIIX import can work is to update the selected word with one of the candidates.


At present, we have no plan at all to import JIIX strokes, as this would be very complicated. The only way to import strokes is as already explained, to convert back the strokes from millimeters to pixels and use the API as done in  the InputController of the uireferenceimplementation.


Best regards,


Olivier


Dear Nicolas,


I answered your other post.


Let us know if you have any other question.


Best regards,


Olivier


1 person likes this

Hi Olivier,

As I mentioned elsewhere, we were able to export stroke data from a Text block to the JIIX format.

I would just like to confirm that it is not possible to import that JIIX stroke data directly somewhere else to then be converted.

The documentation about "Import and export formats" mentions this for importing JIIX to a Text block: "For the time being, JIIX import of text content can only be used to select recognized text word candidates". The part "JIIX import of text content" makes it ambiguous. Does this limitation only apply when importing text, meaning that there is no limitation when importing stroke data, or is "text word candidates" the only option available when importing any kind of JIIX data?

If import of JIIX stroke data is not supported, do you have any plans to support it and any estimated time frame for when it would become available?

Thank you!

Answer

Dear Nicolas,


thank you for the precisons.


To answer your questions, indeed, the only way JIIX import can work is to update the selected word with one of the candidates.


At present, we have no plan at all to import JIIX strokes, as this would be very complicated. The only way to import strokes is as already explained, to convert back the strokes from millimeters to pixels and use the API as done in  the InputController of the uireferenceimplementation.


Best regards,


Olivier