Dear Judy, I general I agree that there are some random components in the process of giving degree.
However if we consider a very simply tests then I don't think that the influence of randomness is significant. Let us consider the test which contains two questions: 1) When did Columbus Discover America? 2) When did the First World War begin? Probability of guessing the answers is very low. What is random in the fact that somebody answers 0%,50%,100% of these question? If somebody learned then he know if somebody didn't learn then he doesn't know. This is quite deterministic relation. Of course in reality there are some random "noises" for example sclerosis, influence of stress etc. However I still don't understand what is random in the fact that somebody remember some data? (if we neglect sclerosis and other problems) Well this is really not a very good example because randomness play an important role in this case. The examples with car (How many parts of the car I will have to remove in order to get something which is not a car?), burning house, (i.e. When burning house become a ruble?) or something similar are much better. I agree that if "Robert is half- German, quarter- French and quarter- Italian," then the numbers 0.5, 0.25 and 0.25 are not probabilities but grades of membership or, equivalently, truth values. However in general I have no idea why I can use t-norms to make some calculation on that numbers. I suspect that this is much more complex problem. Let us consider the definition of the word "car". Well everybody knows what the car is. Is the motorcycle with three wheels a car? There are also other vehicles which are almost a car. How do we know that these objects are almost a car but other objects are certainly not a car? Our brain can estimate that more or less. The question is how in general build a method to estimate the difference between the object and its definition? Unfortunately, I don't know the answer to that question. Maybe some member of this group knows the answer? However I don't see any randomness in this process. Additionally, I suspect that this process is much more complicated than fuzzy logic. I the case of car we have to consider many factors, wheels, engine, material etc. This is multidimensional problem in my opinion. One parameter (i.e. for example membership or whatever we can call that) cannot describe precisely such complicated object like a car. Well in the case of burning w can say for example that the house burn in 50% but I don't thing that this number give precise information about the condition of the house. (What is random in that 50% by the way?) In some cases one number is quite precise information (for example in the case of "fever" temperature give quite good description of the "degree" of "fever"). I can agree that this is some non-probabilistic measure of "fever". Maybe I can call it fuzzy membership function. But why I can make some calculation on that degree using t-norms? What is the relation between t-norms and temperature? What is the relation between t-norms and citizenship of Robert? In other examples there are similar problems. Maybe somebody knows what to do with that? The question is: How well particular object "fit" to its definition? How our brain connect the real object with particular word? Well each child knows the answer to that question, isn't it? When the definitions of the word become invalid? What are the details of that "fitting" process? Regards, Andrzej Pownuk P.S. Some keys fits well to the locks, some keys (with a little damages) cause problems in the process of opening the door. Some keys doesn't fit to the locks at all. In this case the process of fitting is pure deterministic. Additionally I don't know how to describe this fitting using one parameter (I am talking about precise description). And again what is random in that process? > > Andrzej Pownuk wrote: > } Then the degrees at school/university are random? > } Well, I always suspected that :) > > In fact, they are to a large degree. > > I have for several years been interested in modeling academic > advising using Bayes nets. In the US universities, students > have choices; in large universities, the number of options can > be overwhelming, albeit there are constraints imposed by university > and department requirements, and (possibly soft) constraints > imposed by the student's goals and preferences. > > However, there is also the imprecision that comes from varying > perspectives. My understanding of what it means for a student > to do well is that they are earning As or Bs in their classes. > (Our university offers undergraduates 5 grades, A through E, with > E being a failing grade. Because of grade inflation, Cs are > considered bad and Ds quite bad.) > > I have talked to students who perceive that they are doing well > if they are passing their classes with Cs and Ds. > > I have also learned that advisors modulate their advice. For > the good students (B-average and better), they tend to recommend > classes in order to maximize the expected grades. For weaker > students, they tend to recommend classes that the student is > likely to pass. (Some informal comparisons show that these two > strategies lead to different recommendations.) > > So university degrees here are random in two ways: in the > actual classes the students take to fulfill the requirements, and > in the performance the students show in those classes. > > One might argue that the actual grading is also random, dependent > on both the congruence of grading criteria and student skills, and > on the professor's mood. But I won't go there. > ------- End of Forwarded Message
