Comments: |
Pēc būtības Vatsons ir tikai glorificēts meklēšanas algoritms, un priekš tāda tas darbojas visai labi, man domāt. Šāda līmeņa performances pārsniegšana, šķiet, jau ir AGI-complete.
Wasn't the Semantic Net supposed to solve this?
Semantic network is necessary but not sufficient for general intelligence.
I meant not for AGI, but for knowing that Toronto is in Canada.
Given the extensive knowledge base they provided for Watson, it's hardly an error of finding the correct data (although I don't know if the knowledge organisation really formed a semantic network or just some tailored XML-like structure), most probably it's either a failure in inference/decision making caused by failure in language processing, or simply a generic glitch in the software.
| |