News for hackers. Stuff that matters.
Re:Probabilistic Data Structure Showdown: Cuckoo Filters vs.... by kibwen on Monday November 28, 2016 @01:58PM (#13057544)

  > Since it’s possible to hash multiple items to the same 
  > indices, a membership test returns either false or maybe.
As someone who perpetually has trouble keeping the terminology of {false|no false} {positives|negatives} straight, I just wanted to remark that this is a very intuitive and straightforward way of explaining the guarantees of lookup in a Bloom filter.
Re:Probabilistic Data Structure Showdown: Cuckoo Filters vs.... by dexwiz on Monday November 28, 2016 @03:43PM (#13058245)
True/False says if the test was correct. Positive/Negative is if the test passed.

True Positive: Test was correctly passed, hit. False Positive: Test was incorrectly passed, false alarm. True Negative: Test was correctly failed, reject. False Negative: Test was incorrectly failed, miss.

I find the terms to be too technical, so I remember a common situation. Medical tests are often False Positives. Which means the test was positive, but the test was wrong and you are okay (good!). From there is pretty easy to infer the rest.

Infact, that situation leads to a large issue with medical testing at a whole. If you were tested for 100 terrible diseases that you did not have, you would probably get a bunch of False Positives. Medical tests are set up for a high False Positive rate but low False Negative rate, because its less damaging to confirm a diagnosis with follow up testing than to miss a test. But confirming False Positive tests is expensive, along with the initial testing. So doctors try to limit tests to at-risk groups to spend less time on healthy patients' False Positive tests.

Re:Probabilistic Data Structure Showdown: Cuckoo Filters vs.... by jakewins on Monday November 28, 2016 @06:50PM (#13059445)
As an aside, this is how null is defined in many languages as well. Many behaviors of SQL, for instance, are much more intuitive if you remember that SQL sees null as "unknown".

If a boolean value is null in SQL, it conceptually means it could be either true or false - it's a maybe. If you ask "is maybe true equal to maybe true", the answer is "maybe". Likewise, asking SQL 'null = null", it answers "null", instead of true/false.

Re:Probabilistic Data Structure Showdown: Cuckoo Filters vs.... by 443633456 on Tuesday November 29, 2016 @12:03AM (#13060963)
...which is why for determining if something is unknown, you ask SQL...are you indeed NULL? And SQL should say 'true' or 'false'.

    SQL null IS null
Re:Probabilistic Data Structure Showdown: Cuckoo Filters vs.... by gopalv on Monday November 28, 2016 @03:12PM (#13058002)
It's a much better notation because NOT(MAYBE) = MAYBE.

That's hard to explain when you work with binary booleans.

Re:Probabilistic Data Structure Showdown: Cuckoo Filters vs.... by mgunlogson on Monday November 28, 2016 @05:15PM (#13058888)
I try to think of these filters as the opposite of a cache. Bloom/Cuckoo has no false negatives and a cache has no false positives. Besides this they are very similar conceptually and used in the similar contexts.