[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [seul-edu] Eduml proposal (the documennt)
>This look good but why not using ref. to existing <scale> element.
>I took some of your ideas for the mapping.
>For exemple:
>
><scale scaleId="unique scale identifier"> <!-- performance arithmetic -->
> <scaleSchema> a paragraph explaining this marking schema </scaleSchema>
> <scaleMath high="high" low="low" type="relative or absolute">
> mathematical algorithm to use</scaleMath>
> <scaleMatrix> an array of permissible marks if finite </scaleMatrix>
></scale>
>
>
>We can define a common scale based on float value from -100 to 100 for
>exemple. But this is more application specific as Eduml doesn't do any
>computation. In the case of a numeric scale we just need to compute an
>affine transformation using the low and high value. With a discret
>scale we can supply a mapping to the common scale. Scale are quite
>common toward session, so it's ok to defined them outside of the
>session
I'm not so sure that the scale would be the same across classes in a
session - imagine a school which has remedial classes and advanced
classes. The advanced classes may have a different (more difficult)
grading scale... I think we will run into problems unless we allow scales
to be defined for each unique class.
><scales>
> <commonScale>
> <high>100.0</high>
> <low>0.0</low>
> </commonScale>
>
> <scale scaleId="over100">
> <numeric type="???">
><!-- FIXME Should we say if we use int, float, flaten to the 1/2,... ? -->
I think we should always allow floats for numeric marks. People can still
enter whole numbers if they want to.
> <high>100</high>
> <low>0</low>
> <numeric>
> <scale>
<snip>
>Should we add weight to class, course and session ?
I don't know of any reason to do this - I can't think of a situation where
grades for a class would be averaged across a course, for example. I think
assignment and category weights are enough (these are the only items which
affect a _student grade_.)
-Eric
p.s. it's looking good, I think. :)