[gcompris-devel] [GSoc]Games + ReadingWriting

[Alex Burghelea]

On 3/31/2011 11:05 PM, Bruno Coudoin wrote:
> Le mercredi 30 mars 2011 à 04:51 +0300, Alex Burghelea a écrit :
>> Hi,
>>
>> Sorry for the late delay, I had 2 deadlines for this week on my
>> university assignments.
>> I have compiled GCompris and played a little with it.
>>
>> Below you will find my ideas more detailed.
>>
>> On 3/26/2011 6:47 PM, Bruno Coudoin wrote:
>>>> at the dominoes i think that i could combine it with the spelling and
>>>> writing case association.
>>> tack toe, and dominoes
>>>
>>>> Another game i'd like to implement is Dots and Boxes
>>>> (http://jocuri.clopotel.ro/Q7ij8IFL/Dots_and_Boxes/).
>>>> I think this game could have great potential in being integrated with
>>>> the Reading/Writing Category
>>> Hum, Dot and Boxes does not requires any reading / writing skills.
>> Sorry for that one ,  it was meant for the dominoes game.
> Ok, no problem.
>
>> I saw on the Idea Page the different levels proposed for dominoes, and I
>> think that I could implement
>> some levels based on letter ( Case - Sensitive / !Case Sensitive  have
>> to decide which is harder)  and then numbers
>> and finally a harder level with dots (double 6 or double 9 set should be
>> sufficient enough for kids ).
> Great, I like this.
>
>> The main part of the games I'm thinking of implementing the rules of
>> Block / Draw dominoes.
> I don't understand what you mean here.

The Dominoes games are very different , depending of the rules.

Block / Draw dominoes are the most common played. they are played until 
the player about to move can neither draw and or put on the deck a piece 
from his hand . and the deck is only linear , sth like this | 3 4 | 4 6 
| 6 2 |, and the players can only attach to the end ( 3 and 2 ). There 
are versions where you can attach to the sides too .

>> I also want to write another domino game based on the rules from "5s and
>> 3s".
>> http://en.wikipedia.org/wiki/List_of_domino_games#5s_and_3s
> Ouch, it's overly complicated for our target.
>
>> But maybe bended to accommodate simpler math operations ( +/- simple
>> levels | */ : harder levels, or not at all).
> Yes, doing a simpler may work but I am not convince yet. There are some
> research to do to see if it makes sense on a pedagogical aspect.
>
Ok. I will ask a kinder garden  teacher about this.
>> In this part I'm thinking at something interactive like math hints for
>> when the child is making a move that isn't the best one.
>> Let's say that a rule is: "You get points by this formula sum of the
>> dots on your piece and the half you are connecting
>> to."
>> For example: If the kid has in hand 6-3 , 4-2 + other and the deck would
>> end with  4 **** 6  and he chooses
>> to use 4-2 , the sum would be 10<  15 (by playing 6 -3 ). Here is where
>> the math feature is bigger where a child is
>> asked a question like "How much is 3 + 6 + 6?" .
> Well, I thing that's it is way too complex.
>
>> A final game I think of working on, is Connect 4. I would like to
>> implement an interactive tutorial,
>> where the user is presented with some hints that will guide him to
>> victory and
>> maybe even explain (somehow) why the current move it's good/bad.
>> I'm thinking on a solution based on alfa-betha prunning algorithm.
>> I hope that I made myself understood.
> I understand, maybe you could extend this to other games we have. We
> don't provide enough guidance throughout GCompris to help the children
> when he makes mistakes. I am thinking to the awele activity which does
> not show the moves so its hard to understand the rules. Also it could be
> enhanced with helper to show the best moves. By the way, I did manage to
> provide help in the lightsoff activities.

I'll take a look at the lightsoff activities (maybe I'll learn to play 
it :D ).

Sure, expanding to other games, it would be fun.

>> For the reading writing category, I'm thinking that a calligraphic
>> application would be cool.
>> But I don't know how it would be received by the users , because it's
>> limited to using graphic
>> tablets.
> There has already been work on this but it never succeeded. Yes the
> hardware limits the use of this.
>

That was my biggest concern here.

>> I really want to know what's your opinion here.
>>
>> I saw a game that gave me an idea for the Letter Classification
>>
>> http://nobelprize.org/educational/physics/laser/challenge.html
>>
>> At the end of the first level after talking to the police officer, there
>> is a challenge where
>> objects come on a conveyor belt and they have to be categorized.
>> I'm thinking of doing something where the kid has to select letter
>> and put them on several places , so that he could form the words.
> I don't know. I am not a teacher and I have a hard time imagining the
> activity.
>

I'll attach a photo of what I mean. When I was in kinder garden I used 
to play this, and from what I remember it was quite fun :D .

http://swarm.cs.pub.ro/~aburghelea/wordcomp.png 
<http://swarm.cs.pub.ro/%7Eaburghelea/wordcomp.png>

Basically it would be like a spelling bee activity, only the child has 
to use the letter as they come.
I think it will have a bigger impact for hand-eye coordination .

The letters move according to the arrow and the users has to drag and 
drop correct letter in the boxes to complete the word.

>> The dictation of letter i think would be fairly easy to achieve
>> based on a a set of pronounced letter for every language that
>> GCompris supports.  The pronunciation can be used in domino,
>> especially for the level with non case -sensitive letter as hints
>> to match upper to lower-case
> Yes.
>
>> I see the part with word pronunciation trickier
>> because it needs a large database for different languages .
>> A possible solution would be using ogg + Speex . I'm also taking in
>> to consideration some linear prediction ( I had an assignment
>> last year where I had to implement compression and decompression
>> for voice prediction using wav files in octave) .
> Space is not an issue nowadays. The voice dataset are usually optional
> on GNU/Linux. Getting the data right with the propser licence is the
> major issue.
>

Aha. I understand.

>> Do you think it would be enough for a GSoC application (or on the
>> contrary , that I'm biting more that I could chew ) ?
> I don't know, this is hard to tell without a more detailed proposal.
> Also many things you propose are new to me and it would requires more
> time to evaluate them.
>
>> I will also add some bonus task to the proposal like checkers.
> Checkers with an AI ? Would you code it yourself or integrate a third
> party ?

Both with AI , and 2player game.I want to code it myself. In my 
algorithm design course until this year the project was a chess engine, 
so i could take my ideas from some of those assignments. But for 
starters I think I'll use a

third- party and then start writing my own engine to be more suited for 
educational purposes .

I was also thinking at a scrabble game , but I don't know how it will be 
received , being a difficult game. I just had to implement a 
spellchecker, that I think it could be easily adapted for this case. 
*But here i really have my doughs about it will really be usefull.*


>> I will start to write the application for GSoC , and based on your
>> feedback for this mail I will send you a draft by this weeks end.
>>
>> I hope you like my ideas and look forward for your review.
> Fine,
>