by a given algorithm, fuzzy logic inference to be chosen by user from respective base
[13, 14]).
The result of algorithm is a real number in the interval [0,1] or an integer number
in the interval [0,100] - "the exact credibility value" for student knowledge grading.
Optionally, the user based on the received accurate ("clear") grade (CG) can get
"fuzzy" grade (FG), having selected certain rules from a database of fuzzy inference
rules (or through application of "standard" basis) of the following type: "if CG is in the
range [60,65], your score = "solid three".
When using a "uniform" grading it can be possible that one student received 76
points out of 100 corresponding to the grading "good" and the other student - 74 points
corresponding to the grading "satisfactory."
Despite the difference in only 2 points in the knowledge level, students get
different grades, which may be a "cruel joke" in the problem in getting the scholarship.
5-point fuzzy evaluation is less sensitive to minor variations and more in line with
teacher’s psychology (in the opinion of one of the authors of the article - former
student). Membership function (MF) shall be used for fuzzy grading of the following
form (Figure 3):
Figure 3. Membership function
MF form can be explained by the following reasons. When student receives "firm"
excellent grade having solved all tasks or making one minor error, students can get
"firm" good grade when have committed various errors. Fuzzy grade shell is defined
as the closest to a clear grading with the level belonged to it. Depending on MF, let’s
attribute the grading to one of 5 levels.
Thus, a high level (from 0.8 to 1) will correspond to "firm" grading, level (0.5,
0.7) – will correspond to grading "at a stretch." Correlation estimates with an
"intermediate" level (0.7, 0.8) for a particular type depends on the psychosomatic
characteristics of teacher [12].
In determining the total grade in the training course, we may form the rules basis
of grading (or using the standard one built in SMPR system) with output like "you are
almost good", for "good" grading you have to study better such-and-such topic".
- 1844 -