07-04-2020, 11:27 PM
(07-02-2020, 07:39 PM)Mem5 Wrote: Hum. So you are looking at relationships between ESSID and passwords, but you are only interested in passwords created by humans?
IMHO human generated passwords have almost no relationships with ESSID, apart maybe essid==password but you don't need AI for that.
Human passwords are (pet) names, birthdates, etc.
If you're right, the project doesn't make sense after all. I do find cases when a key consists of shuffled word compounds of SSID joined with a symbol or peppered with a simple prefix or suffix. They are not essid==password but typical rules rule them out... and they are rare, I agree. I'd be interested to look deeply into these relations to craft more useful rules.
Besides this, I argue that sometimes there is information in ESSID. If not a plain key, it shifts the probability distribution of favorable candidates to one peak or another. For example, an AP with "TurTelekom_Z4Ry" SSID typically won't use pet names unless the director is a funny man. While APs with SSID "Svetlanochka" have proven to use (pet) names as passwords. Training the distribution of such most likely words could be a good candidate for AI.