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

Do you manually make recommendations for each request or is there some sort of natural language parsing algorithm that you've built? If the latter, then really cool this has potential..


How is it scalable? Also: do you not plan on using ad revenue? As a consumer reports-ish site it seems like ad-fare might damage your cred (read: your entire product).


We show ads to users who aren't signed in. We're experimenting with affiliates, but we're careful to put in affiliate tags after recommendations are generated.

If ads and affiliates become a serious concern for our users, we'll ditch them.


We break up this work into smaller tasks, and use Mechanical Turk for some of the tasks.




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

Search: