Bloom Filter is named after Burton Bloom.
What is bloom filter?
http://blog.csdn.net/v_july_v/article/details/6685894
Define K hash methods.
Given any data, we will have K hashed value.
Set these positions in a bit array as 1.
Check wheter some data is in the set?
Compute K value for this data, and check each position.
It contains "Negative False".
class BloomFilter<T>
{
interface Hash
{
int hash(T t);
}
bit[] bitArray;
Hash[] hashMethods;
void init(Iterator<T> data)
{
// Init a bit array.
bitArray = initBitArray();
// Init k hash methods.
hashMethods = initHashMethods(k);
while (data.hasNext())
{
T t = data.next();
for (Hash hash : hashMethods)
{
int pos = hash.hash(t);
bitArray[pos] = 1;
}
}
}
boolean contains(T t)
{
for (Hash hash : hashMethods)
{
int pos = hash.hash(t);
if (bitArray[pos] == 0)
return false;
}
return true;
}
}Keep Low Fault Rate?
Given max falut rate, Calculate the bit array size?
还有一个比较重要的问题，如 何根据输入元素个数n，确定位数组m的大小及hash函数个数。当hash函数个数k=(ln2)*(m/n)时错误率最小。在错误率不大于E的情况 下，m至少要等于n*lg(1/E)才能表示任意n个元素的集合。但m还应该更大些，因为还要保证bit数组里至少一半为0，则m应 该>=nlg(1/E)*lge 大概就是nlg(1/E)1.44倍(lg表示以2为底的对数)。
