Modelling: Semi-automatic assessment has correct criterion with wrong points

ValentinHerrmann commented 7 months ago

Describe the bug

My assessment criteria for modelling exercises are often worth 0.2, 0.4 or other fractions of full points. The semi-automatic assessment often applies the correct citerium but rounds the fractions to half points.

This is especially problematic because on first sight the assessment suggestion looks correct however leads to a wrong grading.

To Reproduce

Create modelling exercise with a criterium that is worth a non-half fraction of a full point.
Create a least two similar assignments (or more, need to be enough for the semi-automatic assessment to be activated)
Grade one of the assignments manually and use the criterium mentioned at 1.
Proceed to the semi-automatic suggestion and see the wrong result (rounded to half points).

Expected behavior

The same points as in the criterium are expected.

Screenshots

No response

Which version of Artemis are you seeing the problem on?

6.8.3

What browsers are you seeing the problem on?

Chrome

Additional context

No response

Relevant log output

No response

MaximilianJG commented 7 months ago

Hi @LordMuetze, could you elaborate on how I can reproduce this issue? I have been trying on TS2 with the MaxGlasmacherTestCourse using the WrongPointsAssessmentExercise but can't seem to get manual assesment to work. A little bit more detail would be great, thanks!

ValentinHerrmann commented 7 months ago

Hi @MaximilianJG, unfortunately I do not know how to access your test course. Maybe you can give me direct access so I can have a look at the exercise you created? I will send you more details on the actual exercise and students via Slack.

ValentinHerrmann commented 7 months ago

Today an additional bug occurred (example screenshot below, explanation following):

All automatically applied criterions are displayed with 0.5 points (which is wrong and should be 0.2 points).
6 criterions (which should be 0.2 points each and actually are 0.5 each) are applied however the sum is displayed as 1.8 points.
When reapplying the correct criterion the points of the single criterion are adjusted correctly. However the overall sum ist sometimes adjusted and sometimes not. Unfortunately I could not identify a pattern of when it works and when it doesn't, yet.

ValentinHerrmann commented 7 months ago

I just recognized that at most occurrences of the second part of the bug the following scheme applies:

0.2 points criterions are always displayed as 0.5 points
When summing up the overall points only on the first application of the criterion 0.5 points are used. For the other applications of a criterion - despite 0.5 points being displayed - the correct score is used in the overall sum.

As I have already overwritten most of the buggy scores I cannot very this assumption.

ls1intum / Artemis