Inadequate Thesaurus in Writer

[Németh László]

Hi Roger, hi all (CCed to the l10n list, too),

Unfortunately, this is a regression of the thesaurus, related to the
carelessly updated English spelling dictionaries. I filed a bug for the
problem now, suggesting also a quick solution:

https://bugs.documentfoundation.org/show_bug.cgi?id=167649

Thanks for your bug report!

More information:

The word “peered”, like other affixed word forms, is missing from WordNet,
which is the source of the English thesaurus of LibreOffice. LibreOffice
was able to handle this problem by stemming and affixation, using spelling
dictionaries extended with morphological data, which allowed to suggest
affixed forms, e.g. here: “looked” for “peered” (but not
“gazed”, “glanced”, “peeked” etc. or “examined”, “inspected” suggested by
chatGPT, which contains much more data, including synonyms, than WordNet).

As the author of the morphological extension of the English dictionaries,
also as the author of the related code parts in the spell checker and
thesaurus libraries, my aim was to create a reference implementation for
other languages. In fact, this was a voluntary contribution, by-product of
developing the Hungarian thesaurus, supported by the FSF.hu Foundation,
Hungary.

I was sorry that my work was deleted by mistake (kept only by the git
repository), but this is not unprecedented. For example, the other thing
that was similarly deleted, and which I regret very much, is the old style
manager interface that allows copying single paragraph etc. styles between
templates or documents.

For l10n, I repeated some information about developing thesauri with
stemming and affixation in other languages, because it seems, the original
archive of the OpenOffice.org Lingucomponent mailing list is not very
accessible:

https://bugs.documentfoundation.org/show_bug.cgi?id=167649#c2

Thank you for the bug report again! Hopefully thanks to this, the reported
problem will be fixed soon, and maybe we'll see further improvements in
English and in other languages supported by LibreOffice!

Best regards,
László


Roger Squier <rogersquier at gmail.com> ezt írta (időpont: 2025. júl. 19.,
Szo, 17:26):

> I hope I am sending this email to the right place. It is not about a
> 'bug'; it is about improving the quality of the Thesaurus in Writer
> 25.2.5.2.
>
> Having migrated from MS Word to Libre Office Writer, I’ve found the
> English Thesaurus included in Writer hopelessly inadequate.
>
> Would it be possible for the developers create or find a better Thesaurus?
>
> Just as an example, a search for a synonym for ‘peered'’ in Writer
> produced the result shown in the attached file. The Thesaurus appears to
> not understand English grammar. 'Peered' is a past participle of the verb
> To Peer, e.g. To Look; it is not a noun. This result is similar to some of
> the other results I get when I want to find a synonym for a word in Writer.
> I have even wondered if Writer Thesaurus is going by the sound of a word
> and not the grammar.
>
> See for example File Hippo: *https://www.wordhippo.com/what-is/another-word-for/peered.html
> <https://www.wordhippo.com/what-is/another-word-for/peered.html>*
>
> Surely Writer Thesaurus can be improved significantly.
>
> Regards
>
> Rwsjr
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/libreoffice/attachments/20250723/47ed05d4/attachment.htm>