gaika

39 ms·

391.
by gaika 19y ago
Popular stuff is easy to predict, it is the users with little feedback and less popular movies that make all the difference. Digg doesn't need a recommender because it is all about popularity, and if you do not like it - do not go there. I
392.
by gaika 19y ago
Any chance to see it used here? Another menu at the top: recommended?
393.
Mathematica's Google Aptitude(mathworld.wolfram.com)
24 points by gaika 19y ago|3 comments
394.
Can a spam filter play chess?(dbacl.sourceforge.net)
1 points by gaika 19y ago|0 comments
395.
by gaika 19y ago
I can be used for good. Netflix comes to mind. They say there are only 800 people in the world, the rest are androids.
396.
by gaika 19y ago
Yes. But it happened on HN too: I watch the logs all the time and cannot stop myself from jumping into discussion once a link to my site is posted somewhere, making it seem like I'm behind the original post. Probably should just stop doing
397.
Ask HN: being accused of running sockpuppets, what would you do?
1 points by gaika 19y ago|3 comments
398.
by gaika 19y ago
We just lack the courage to acknowledge our own lack of courage. Ironic.
399.
by gaika 19y ago
Unlike reddit or news.yc, jaanix has a recommender that is based on implicit trust between the users. Sock puppets destroy this trust.
400.
by gaika 19y ago
Somehow I still feel cheated. I'm working on a social news site, and we do not use ghost accounts or fake stats. People still suspect us doing it anyway :( Now I know - they have a good reason to suspect everybody .
401.
by gaika 19y ago
The one that is caused by aliens? ( http://www.crichton-official.com/speech-alienscauseglobalwar... ) Learn, ignore politics, care about real issues that affect you and people that you know.
402.
by gaika 19y ago
Go for it, it is even more fun that way.
403.
by gaika 19y ago
Would you, as a human, prefer a model that gives you accurate predictions with less data, or a model (that is simpler) but needs more data to be accurate? How is an algorithm any different?
404.
by gaika 19y ago
Not true - from current leaders of netflix prize: Our experiments clearly show that once you have strong CF models, such extra data is redundant and cannot improve accuracy on the Netflix dataset. http://glinden.blogspot.com/2008/03/using-
405.
by gaika 19y ago
"I want the process of reading news to be like an old friend telling me gossip that he will think I'm interested in hearing." Do you know such sites? I'd love to read like this too.
406.
by gaika 19y ago
I wonder if it can be used as an anti-troll weapon in other discussion forums... Should it be added to the news.YC guidelines too?
407.
by gaika 19y ago
A plan for spam is born :) Yes, I can tell you in advance, they are distinguishable - these are the stories that YOU will never vote up. And one can predict how you will vote on the stories with a pretty good accuracy with statistical facto
408.
by gaika 19y ago
Ironically, this is a typical reddit post too. And why do you think that posts suggesting different strategies to combat noise are not worthy your attention? This is #1 problem for any news site, and this one of the best communities to disc
409.
by gaika 19y ago
from http://mauitian.livejournal.com/11166.html The last thing you need to know about girls has to do with love. Even just thinking about girls is likely to cause you more emotional anguish than anything else in your life. And it's not li
410.
by gaika 19y ago
she was trained, took ~5 years - http://www.youtube.com/watch?v=8FGEJKQzaMA
411.
by gaika 19y ago
Cool! I like it, we need even more sites like this! Digg, reddit, mixx, stumbleupon, and hacker news is not enough in this crazy world of abundance of choices. How about adding some real-time tuning options?
412.
by gaika 19y ago
Do you know that with a proper recommendation system you can vote both the user AND the story at the same time?
413.
by gaika 19y ago
Yeah, simple trust metric would be very nice here. Give more weight to people who vote like you and everything will be ranked as if you did it.
414.
by gaika 19y ago
It is possible, and we have done it on jaanix. Take a look, there are no surprises. You have a guest id with no questions asked that lets do everything on the site. You can keep it forever if you come back at least once every two weeks. You
415.
by gaika 19y ago
It is using machine learning approach to estimate what the impact of your feedback should be. That way it is depends on context (only topics / tags right now) and is gaming proof (i.e. you can give as much feedback as you want, that would
416.
by gaika 19y ago
Take a look at the speed of adoption - http://www.indeed.com/jobtrends?q=ruby%2Cpython%2Cperl%2C+ja...
417.
by gaika 19y ago
Trust and reputation are always in context, current karma system ignores it. But once you add context visualizing and interpreting karma becomes hard. Take a look of how we're trying to deal with this problem (link in my profile, sick of be
418.
Is Ruby surpassing Python, Perl any time soon?(indeed.com)
8 points by gaika 19y ago|13 comments
419.
by gaika 19y ago
If you're asking about recommender behind jaanix you can contact me on skype - krotta. Tim2 wrote his own recommender for his site, we're not affiliated.
420.
by gaika 19y ago
A few more instances like that and news.yc might become another reddit. Are there any plans to have better karma scoring system? More merit based and less popularity contest?

More ›