iink SDK on Device

Answered

How is the range of styling decided when underlining text?

We are trying to understand the rules for applying styling when we draw an underline under some text with a stylus and the iink SDK. We are testing with text in Japanese.

It appears that if we underline only some part of a word, sometimes the styling will be applied to single characters and sometimes to the whole word but we don't quite understand the conditions for each type of bevaviour.

Our current guess is that the SDK always attempts to stylethe whole word if enough of the word is highlighted, ignores the underline if not enough is underlined, but sometimes fails to recognize a group of characters as being a single word in which case it will style only the characters over the underline instead of the whole word.

Is the above correct? Are there other conditions? Or are the conditions different from what we guessed?

Thank you!



Best Answer

Dear Nicolas,

currently our algorithm works as follows: If more than half of a word is underlined, then we underline the whole word, otherwise we don't. This normally applies to all languages

In the specific case of japanese, sometimes words are not prolery segmented and are recognized as characters, which may be the cause of your behavior. Normally, our technology is pretty robut to this: do you have custom configurations that would make words are recognized as characters?

Best regards,

Olivier


Answer

Dear Nicolas,

currently our algorithm works as follows: If more than half of a word is underlined, then we underline the whole word, otherwise we don't. This normally applies to all languages

In the specific case of japanese, sometimes words are not prolery segmented and are recognized as characters, which may be the cause of your behavior. Normally, our technology is pretty robut to this: do you have custom configurations that would make words are recognized as characters?

Best regards,

Olivier

Thank you Olivier.

We don't have a custom configuration for word recognition. We use the Japanese language resources available for download from your website.

As a follow up to the question, our client would like to know the source of the data you used to build the Japanese words and characters databases, if possible.

Dear Nicolas,

currently, we do not communicate much regarding our sources.

We use varied sources, like written documents, news articles, messaging and spoken.

Best regards,

Olivier