c******n 发帖数: 4965 | 1 http://blog.notdot.net/2012/09/Dam-Cool-Algorithms-Cardinality-
a very often-asked question, the above is the defacto standard method used
in many implementations today (including redis) , it's the result of 20
years of research. it's really stupid and unfair to bring this seemingly
simple question to interview, I bet most of the interviewers don't know the
above at all. |
j********r 发帖数: 127 | |
b**********5 发帖数: 7881 | 3 这个我面dataminr的时候, 被问到, 然后给了个bloom filter, 然后就被拒了
the
【在 c******n 的大作中提到】 : http://blog.notdot.net/2012/09/Dam-Cool-Algorithms-Cardinality- : a very often-asked question, the above is the defacto standard method used : in many implementations today (including redis) , it's the result of 20 : years of research. it's really stupid and unfair to bring this seemingly : simple question to interview, I bet most of the interviewers don't know the : above at all.
|
w**z 发帖数: 8232 | 4 和你说过了,用linear counter,除了刷题,拜托也学学其他的。
【在 b**********5 的大作中提到】 : 这个我面dataminr的时候, 被问到, 然后给了个bloom filter, 然后就被拒了 : : the
|
w**z 发帖数: 8232 | 5 we use linear counter to count uniques. there is open source implementation.
http://highscalability.com/blog/2012/4/5/big-data-counting-how-
the
【在 c******n 的大作中提到】 : http://blog.notdot.net/2012/09/Dam-Cool-Algorithms-Cardinality- : a very often-asked question, the above is the defacto standard method used : in many implementations today (including redis) , it's the result of 20 : years of research. it's really stupid and unfair to bring this seemingly : simple question to interview, I bet most of the interviewers don't know the : above at all.
|