Has bloom filters a higher probability of collide when strings are similar

Question

Because if it is the case, this really suits my need.

I'm trying to discover popular searches on a website. For this, I'm using TopK algorithm which is based on Bloom Filter hashing.

I don't want "Hello world" and "hello world" to be count twice. So if it is a collision, that would be really appropriate for my use case.

Practically, I'm using this implementation.

Practically, I can run String similarity prior to adding to TopK bucket, like for any new string item, if it is similar to at least one String in the bucket, then convert it to the one in the bucket before adding. But this would be stacking same logics (in case it could be achieved directly with Bloom filters).

Answer 1

It depends on the hash function but in general no, because standard hash functions are designed to avoid that similar objects have similar hash codes. [edited, see comment]

For your use case you should probably use a custom hash function which makes it more likely for two strings to have the same code. A one-hot encoding of the characters or characters bigrams in the string might work.