Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I actually built my own version of Watson based on this idea - jeopardy questions are often google-able/searchable on wikipedia. It was pretty easy to build out - but it only gets 80% of the way there.

The last 20% is the hardest - and it's why Watson is so impressive (even though even Watson is probably only at 90%)



Did you every see that HN Post about QUANTA [1]? Do you think it would cover the last 20%?

[1] https://news.ycombinator.com/item?id=8298887




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: