This is the code that was used for the benchmark in the paper "Approaching Optimal Duplicate Detection in a Sliding Window", to be published in COCOON'2020.

Many parts of this code are re-used from a previous benchmark, available at https://bitbucket.org/team_qht/qht/src/master/

Dataset

This benchmark only uses artifical datasets, for reasons explained in the paper. The streams are randomly generated on the spot.

Requirements

The code uses boost (sudo apt-get install libboost-dev). Compilation is done via the Makefile (type make), generated binary is build/qht. The compiler must be C++17 compliant.

Notes

• Hash functions return size_t hashes, which are 64 bits long (or 32 on a 32-bits machine). This code may underperform on 32-bits machine.
• Other filters than QHT and its variants have not been space optimized. Due to specific implementation the parameter choice may be a bit restrained.