GPU-Based Minwise Hashing

Minwise hashing is a standard technique for efficient set similarity estimation in the context of search. The recent work of b-bit minwise hashing provided a substantial improvement by storing only the lowest b bits of each hashed value. Both minwise hashing and b-bit minwise hashing require an expensive preprocessing step for applying k (e.g., k = 500) permutations on the entire data in order to compute k minimal values as the hashed data. In this paper, we developed a parallelization scheme using GPUs, which reduced the processing time by a factor of 20-80x. Reducing the preprocessing time is highly beneficial in practice, for example, for duplicate web page detection (where minwise hashing is a major step in the crawling pipeline) or for increasing the testing speed of online classifiers (when the test data are not preprocessed).

p565.pdf
PDF file

In  21st International World Wide Web Conference

Publisher  Association for Computing Machinery, Inc.

Details

TypeInproceedings
> Publications > GPU-Based Minwise Hashing