Count-Min-Log sketch: Approximately counting with approximate counters
This work addresses a specific deficiency in approximate counting for large-scale processing, particularly beneficial for text-mining tasks, but it is incremental as it builds on existing Count-Min Sketch variants.
The paper tackles the problem of high relative error for low-frequency events in the Count-Min Sketch algorithm, proposing the Count-Min-Log sketch that uses logarithm-based approximate counters to improve average relative error while maintaining constant memory footprint.
Count-Min Sketch is a widely adopted algorithm for approximate event counting in large scale processing. However, the original version of the Count-Min-Sketch (CMS) suffers of some deficiences, especially if one is interested by the low-frequency items, such as in text-mining related tasks. Several variants of CMS have been proposed to compensate for the high relative error for low-frequency events, but the proposed solutions tend to correct the errors instead of preventing them. In this paper, we propose the Count-Min-Log sketch, which uses logarithm-based, approximate counters instead of linear counters to improve the average relative error of CMS at constant memory footprint.