Problem with matching game?

[delm13]

Hi there,

I think there is a problem with the matching game, but not sure.
I am picking two lessons to work with.  One lesson has about 1200
right, the other lesson has about 1600 right.  The problem is
almost all the words come out of the 1600 lesson?  Not the 1200
lesson.  Shouldn't the game pick more words from the lesson that
has less total right?

I'm trying one right now that has 491 right another that has 1078
right.  I would think the game would pick more words from the 491
lesson, correct?  Hmm, this time it did pick more from the 491 lesson.

I'll try the one with 1324 and 2045, yep, didn't work this time, out
of 24 tries, only 4 came from the 1324 lesson and 20 came from the
2045 lesson.....

So, I think something is wrong with how the matching game picks
words from lessons.

Thanks, Del

[Misha (Internet Polyglot)]

Hi delm13.

I tried to reproduce the problem that you described but I couldn't. Here is what I did:
- I went to the database and exported all data of your game results to a text file
- I created a user with your name in our development database and imported the results of your games to our internal database (so that my experiments do not create problems on your account)
- I started playing on our dev site (I logged in with your name). I selected English-Russian lessons "Animals" (you have 607 rights on it at the time of export) and "Buildings" (1727 rights)
- I played both Guessing Game and Matching Game and in both cases I had about 80-90% of words from lesson Animals - which is what I expected.

So it seems that in this experiment everything works as expected. If you think that I missed something in the approach - let me know.

I appreciate much that you point it out and would like to stomp the problem since it seems quite serious. Is it still showing words from better known lessons on your side as opposed to less known?

Misha

[Misha (Internet Polyglot)]

Hi delm13,

Any update on this problem? Is it still existent on your side?

Misha

[delm13]

Right now I am trying Animals and Cars and almost all the words
are coming from the Animals one??  Not sure why?

OH!!!  I see what is happening.....  Look:
Animals .... 52 words each about 14 times = 728
Car        ... 25 words each about 16 times = 400

So, I think the computer should pick more words from the Car lesson
as there are only 400 points, where Animals has 728...

BUT.. the computer is picking them from Animals, because they only
have 14 while Car has 16 per word in the lesson.

***So, it is working correctly I would say***

I thought it was picking words by the total "score" of the lesson, but
it picks the words based on the score of the individual words in the
lesson.

P.S. One way to show this would be to add one more item on the
lesson page.  Right now at the main lesson page the bottom of each
lesson says: "• Author: polyglot    • Right: 1781.0   • Wrong: 9.0"

What I would do is add one more entry at the end called • Average: ??"
The average would be the number right divided by the number of
words in the lesson.  Then you could see that even though a lesson has
a high score, it has a low average because there are many words.

[Misha (Internet Polyglot)]

What I would do is add one more entry at the end called • Average: ??"
The average would be the number right divided by the number of
words in the lesson.  Then you could see that even though a lesson has
a high score, it has a low average because there are many words.

Good idea. I have put it to the list of our improvements.

Misha

[Sultan]

P.S. One way to show this would be to add one more item on the
lesson page.  Right now at the main lesson page the bottom of each
lesson says: "• Author: polyglot    • Right: 1781.0   • Wrong: 9.0"

What I would do is add one more entry at the end called • Average: ??"
The average would be the number right divided by the number of
words in the lesson. 

Hi delm13,

I am a frequent visitor of InternetPolyglot.com and I appreciate your idea of showing 'Average: ??' with the Right / Wrong numbers. It sounds very useful. However, I think the suggested calculation would be quite confusing to the user. Here's why I think so:
 
Say, you have 1718 Right guesses for a lesson having 52 words. The Average would come to 1718/25 = 34.25.

Now, what would that number (34.25) mean to a language learner like me (who is not very good at math :p) ?  That I guessed an average of 34.25 words right?

On the other hand, if it was something like 'Average: 34%' , it might suggest that 34% of my tries were right. OR, it could even suggest that I guessed 1718 words right which is 34% of the total words in the lesson! Clearly, that's not the correct info.

So, I am already confused about this - forgive me if I confused you too   I would appreciate if you give some examples to help me understand it before the Polyglot's team implements this feature.

cheers!

[Misha (Internet Polyglot)]

Thanks Sultan.

delm13, we almost deployed the new version with averages but it seems that it can be confusing for new users. I understand that for you as an advanced user it will give more information but we need to think about all users, including those who just came to the site. For them too much information can be overwhelming.

Let's think what we can do to satisfy both of user groups.

Misha

[delm13]

Yes, the name "average" is too generic... i.e. the average of WHAT?

After doing some thinking, what it is saying, is how many times you
have "said" or 'tested" all the words of the list.  In other words, if you
have gone through the list 10 times, the average is 10. (about anyways!)

(also, I would not put decimal numbers.. just truncate the answer)
(so, 10.2 --> 10;  12.9 --> 12 see??)

What we need is a more accurate name for 'average', some suggestions
might be, "Times list completed", "Lesson completed", Lesson views", etc.
(I used the word lesson as you used that term on the home page)
There might be other good names, or better ones, but these were the
ones I came up with off hand.

[Sultan]

Thanks, delm13 and Misha, for sharing your thoughts on this.

As a learner, knowing how many times I completed or repeated a lesson might not be as useful an info for me as knowing how well I know the words in that lesson. For example if I know that I repeated a lesson 12 times and I can still not guess half of the words right, it might not motivate me much. However, if I know that my skill level with that lesson is 5 out of 10, or 50%, I could have a sense of achievement.

So, why not just call this score something like 'Skill level: 5' or 'Skill: 50%' This number would then be calculated as Total_right/total_words * 100 (or * 10 to scale it down to 10). And, of course, we should round the number rather than show decimals.

My two cents!

[delm13]

Hmm, yes, that could be helpful.  That is measuring something
different though, then what I was looking for, which is how many
times (on average) I've reviewed each word correctly.

However, there is no reason both couldn't be added.  Something
like "Lesson views: X,  Skill level: X%".

But , I wouldn't use the word Skill level, I would use something
like Grade, Score, %Right, etc.

[Sultan]

Thanks, delm13,

I agree with both of your suggestions.

Sultan