You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Since the grammar checker is also doing spelling, it should also use the speller API's of LibreOffice. This has several benefits:
we can do away with the standard speller, which means less downloads/installations
with no external speller, we don't risk the speller creating noise / false alarms or in general behaving worse than the grammar checker+speller; this is true especially when it comes to tokenisation (see example below)
unless anounced as a speller, LO will not indicate that there are proofing tools unless also the speller is installed, thus we need to announce the grammar checker as speller to make sure that users are aware that there are proofing tools installed
speller suggestions coming from the grammar checker are usually much better, and will thus create an improved user experience:
some suggestions are removed by cg rules based on context
for all misspellings being recognised and analysed with an error tag, we can generate only the expected suggestion, nothing else
As an example of issues we get away with using this approach, consider this sentence:
Sámedikki kultur- ja valáštallanstipeanddat
The regular speller will mistokenise kultur- and only see kultur, which it does not recognise and thus gives a red underline. The grammar checker tokeniser, on the other hand, does it correct, recognises the full form kultur-, and sees that it is correct. No underline added.
The grammar checker should use the API such that only errors tagged with &typos are provided to the speller API (offsets for redlining, list of suggestions).
The text was updated successfully, but these errors were encountered:
Since the grammar checker is also doing spelling, it should also use the speller API's of LibreOffice. This has several benefits:
As an example of issues we get away with using this approach, consider this sentence:
The regular speller will mistokenise
kultur-
and only seekultur
, which it does not recognise and thus gives a red underline. The grammar checker tokeniser, on the other hand, does it correct, recognises the full formkultur-
, and sees that it is correct. No underline added.The grammar checker should use the API such that only errors tagged with
&typos
are provided to the speller API (offsets for redlining, list of suggestions).The text was updated successfully, but these errors were encountered: