[GCompris] New grammar activities with internal dataset

Johnny Jazeix jazeix at gmail.com
Thu Dec 14 07:13:37 GMT 2023


Hi Franklin,


Le jeu. 14 déc. 2023 à 04:27, Franklin Weng <franklin at goodhorse.idv.tw> a
écrit :

> Hi,
>
> A few suggestions -
> 1. If there are more than one nouns/verbs/... to be selected, maybe add a
> hint if users choose less than the answers.  I was confused when I just
> selected one noun and it just told me it was wrong, nothing else.
> 2. Some words are too close. (See screenshot:
> https://nextcloud.slat.org/index.php/s/RdkxB2jL9TM2E3A )
>

Thank you, we'll take a look at those 2 points.


> 3. I wonder how the languages other than en are applied.  Though I can
> create json files for Traditional Chinese as well, what I'm thinking about
> is to let kids(users) learn English/French/Spanish/... etc.  I created -en
> json files, copied them to -it files, put some obvious differences like to
> modify Find to ITFind.  Then select language to Italiano.  It still showed
> Find instead of ITFind.  I even deleted -en files, leaving -it files only.
> It still showed Find instead of ITFind.  I guess it didn't work the way I
> thought it is.
>
>
The build folder needs to be cleaned, run cmake again to make sure the new
file is taken in account due to the way we handle resource (with a glob
instead of writing files one by one and new files are not discovered this
way).
The other solution is you don't recompile GCompris is to add the file
https://invent.kde.org/education/gcompris/-/blob/master/src/activities/grammar_analysis/examples/grammar_analysis-xx.json
(no rename on the -xx, keep it this way) to
$HOME/.local/share/GCompris/grammar_analysis-xx.json and use this one to
write your dataset. Then, if you send it to me (no need to add the whole
list), I'll push it i git with the good filename.

Regarding languages with different "rules", the developer of the activity
tried to handled several rules as explained below such as right to left
languages, languages with no spaces... If there are more rules for
Traditional Chinese that we need to follow, please tell us. If you can
create a dataset (even small) to point us all the differences so we can
make the activity working for ths language, it would be great!


>
> Franklin
>
>
Cheers,
Johnny


> Johnny Jazeix <jazeix at gmail.com> 於 2023年7月24日 週一 下午11:57寫道:
>
>> Hi,
>>
>> We have 2 new grammar activities in GCompris (thank you Bruno for
>> these two new activities, they will be appreciated). The levels for each
>> language are in separate .json files, to allow definitions of
>> language-specific grammar classes and exercises.
>>
>> The 2 activities are very similar, the only difference is that one
>> (grammar_classes) only asks for a single grammatical class per level,
>> while the other (grammar_analysis) asks for several grammatical classes
>> per level.
>>
>> We have made special development packages, with a special translator
>> option activated to let you easily create and test localized json files.
>>
>> Linux (64bit):
>> https://gcompris.net/beta-download/gcompris-qt-4.0-dev-grammar-Linux64.sh
>> Windows (64bit):
>> https://gcompris.net/beta-download/gcompris-qt-4.0-dev-grammar-win64.exe
>> Android (32bit):
>>
>> https://gcompris.net/beta-download/GCompris-Android-release-signed-aligned-armeabi-v7a-dl-4.0-dev-grammar.apk
>>
>> To create json files for your languages, we have a very detailed README
>> (
>> https://invent.kde.org/education/gcompris/-/blob/master/src/activities/grammar_analysis/README.md
>> ),
>> but as it can be long to read, we tried to to summarize the most
>> important points below.
>>
>> - First, copy the base example files
>> (
>> https://invent.kde.org/education/gcompris/-/blob/master/src/activities/grammar_analysis/examples/grammar_analysis-xx.json
>> and
>>
>> https://invent.kde.org/education/gcompris/-/blob/master/src/activities/grammar_analysis/examples/grammar_classes-xx.json
>> )
>> to the following path depending on your operating system: on Linux, it
>> is $HOME/.local/share/GCompris/, on Windows it is
>> C:\Users\Username\AppData\Local\gcompris\, on Android it is
>> /Android/data/net.gcompris.full/files/GCompris/.
>>
>> - In each file, start by editing the "syntax" section at the end. It
>> must contain one entry for each of the grammatical classes that will be
>> included in the sentences. For each entry, you'll need to set a "code"
>> (a short identifier to refer to this class in the other sections), a
>> "wordClass" (the full name of the class which will be displayed in the
>> activity), and an "image" name for the symbol which will be associated
>> to this class in the interface of the activity. There are currently 10
>> images available, named "1.svg" to "10.svg". If you really need more,
>> tell us and we'll add some.
>>
>> - Note that we only expect main grammatical classes, not sub-classes
>> (for example in English, we only have one "pronoun" entry, not one for
>> each type of pronoun).
>>
>> - The next section to edit is "dataset", in the middle of the files. It
>> contains some groups of sentences which can be reused in different
>> levels. It is also possible to create one sentence group per level, but
>> the possibility to reuse sentence groups in different levels can reduce
>> the amount of work for you and optimize the size of the packages. There
>> are 3 groups by default (named "small", "medium" and "large"), but you
>> can add as many groups as you need, and name them as you want.
>>
>> - Each entry in a sentence group has a "sentence" value (to write the
>> actual sentence to display) and an "answer" value which must contain the
>> list of grammatical classes corresponding to the words in the sentence,
>> in the same order. If you want, you can add more than one space between
>> identifiers to align them visually with the words. Example:
>>        { "sentence": "The flower is very pretty."
>>        , "answer":   "dtm nou   vrb adv adj"
>>        }
>>
>> - If you need you can group several words to match a single class by
>> using parenthesis. The only limitation is that the grouped words must be
>> next to each-others. Example:
>>        { "sentence": "I (am reading) the news."
>>        , "answer":   "prn   vrb      dtm nou"
>>        }
>>
>> - If your language allows to agglutinate words together, you can specify
>> more than one class for a word by using +. Example:
>>        { "sentence": "ശേഷം ആനയോടി."
>>        , "answer":   "avyayam naamam+kriya"
>>        }
>>
>> - As a general advice, please try to make relatively small/medium sized
>> sentences, and using small/medium sized words, else the sentences might
>> overflow the allocated screen space.
>>
>> - The last section, at the beginning of the file, is to create "levels".
>> Each level contains 4 values:
>> "objective": the description displayed in the activity.
>> "difficulty": the difficulty level (not used for now).
>> "exercise": the name of the sentence group to use.
>> "goal": the list of grammatical classes to identify in this level.
>>
>> In the provided development packages, as the translationMode is
>> activated, all the sentences in the exercise groups will be displayed
>> for each level, in the same order as they are written in the json file.
>> Else for users, sentences will be filtered to use only those containing
>> at least one of the requested classes (in the goal) in the level, and in
>> random order.
>>
>> If you need more info, please read the README, and if you still have
>> some doubts you can ask us.
>>
>> When your files are ready, you can send them directly to me or Timothée
>> and we will add them to the activities if you don't have direct commit
>> access.
>>
>> These activities will be in our next release that we usually plan to
>> do between December/January.
>>
>> Cheers,
>>
>> Timothée & Johnny
>>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-i18n-doc/attachments/20231214/6d3c8af7/attachment-0001.htm>


More information about the kde-i18n-doc mailing list