3 ms·

As the logic dealing with {0, 1} is named after its founder George Boole, a boolean or bool for short. I'd suggest we call a fuzzy truth value in the interval

by j-pb 6y ago

As the logic dealing with {0, 1} is named after its founder George Boole, a boolean or bool for short. I'd suggest we call a fuzzy truth value in the interval of [0,1] after its founder Lotfi A. Zadeh, a zadehan or zade for short. Edit: Fixed Bool to Boole thanks to globular-toast, my internal syntax checker must have auto corrected that one ;)

globular-toast 6y ago
*George Boole. I like this idea, although I don't think English speakers would know how to pronounce "zadehan" by default. "Zadean" might be better. I don't think it will catch on, though, because "boolean" is really easy to say but "zadean" isn't.
eps 6y ago
"Zad" means "rear" or "behind" in slavic languages and it's closely related to "butt" and "ass". So the appeal of your suggestion, as clever as it is, is definitely not universal :)
dragonwriter 6y ago
I'd only use Zadehan if it not only was a value on the interval [0, 1] but a fuzzy membership value on that interval to which the Zadeh fuzzy logic operators apply. If it was a probability value to which Bayesian operators apply, Zadehan would be a singularly inappropriate name for the type. Types aren't just ranges but they also define the valid operations (whether syntactically functions, methods, or operators) on the type.
- j-pb 6y ago
  Sure, we can have both a zadeh, and a bayes ;) I'm not sure if I find Koskos argument convincing, however one could also argue that a bayes is a subtype of a zadeh ;) http://sipi.usc.edu/~kosko/Fuzziness_Vs_Probability.pdf http://sipi.usc.edu/~kosko/Fuzziness_Vs_Probability.pdf
  - dragonwriter 6y ago
    Bayesian probability is, from a certain perspective, a fuzzy logic but the operators of Bayesian logic are not the operators Zadeh’s fuzzy logic. Since types define operators as well as values, I'd argue that a “bayes(ian)” type should be distinct from a “zadeh(an)” type, even if they might usefully have a common “fuzzy_membership” supertype.

by j-pb 6y ago

globular-toast 6y ago

*George Boole. I like this idea, although I don't think English speakers would know how to pronounce "zadehan" by default. "Zadean" might be better. I don't think it will catch on, though, because "boolean" is really easy to say but "zadean" isn't.

eps 6y ago

"Zad" means "rear" or "behind" in slavic languages and it's closely related to "butt" and "ass". So the appeal of your suggestion, as clever as it is, is definitely not universal :)

dragonwriter 6y ago

I'd only use Zadehan if it not only was a value on the interval [0, 1] but a fuzzy membership value on that interval to which the Zadeh fuzzy logic operators apply. If it was a probability value to which Bayesian operators apply, Zadehan would be a singularly inappropriate name for the type. Types aren't just ranges but they also define the valid operations (whether syntactically functions, methods, or operators) on the type.

j-pb 6y ago
Sure, we can have both a zadeh, and a bayes ;) I'm not sure if I find Koskos argument convincing, however one could also argue that a bayes is a subtype of a zadeh ;) http://sipi.usc.edu/~kosko/Fuzziness_Vs_Probability.pdf http://sipi.usc.edu/~kosko/Fuzziness_Vs_Probability.pdf
- dragonwriter 6y ago
  Bayesian probability is, from a certain perspective, a fuzzy logic but the operators of Bayesian logic are not the operators Zadeh’s fuzzy logic. Since types define operators as well as values, I'd argue that a “bayes(ian)” type should be distinct from a “zadeh(an)” type, even if they might usefully have a common “fuzzy_membership” supertype.