I downloaded adroll/cantor to experiment getting hash counts of dataset intersections. It uses Hyperloglog (HLL) + minhash but I find the test cases insufficient. I need to know if ALL intersection calls use minhash even in a minimal way. I found no blogs showing test cases and their feedback. The documentation does not provide that fine-grained detail. Anyone out there experienced with adroll/cantor? - best wishes
I have i.e. the K-parameter. Since that message, I have become more experienced and obtained successful results. This code does not use inclusion/exclusion but minhash throughout. I managed to get good results but there is a time factor with increased space.
Your stuff has nothing to do with my question. Please get out of this forum and seek help.
The Adroll/Cantor has been sussed out some time ago. Adroll/Cantor does not use the inclusion-exclusion principle but rather (k)minhash algo. The more (and more time) and the better the intersections. It's a good program for what it does.
- Badges
- Science topic