This project is about building a new cache replacement policy, one that improves on older methods like LRU and LFU. Those older methods assume every piece of data is either clearly useful or clearly not. In reality a lot of data falls in between, it might be contradictory (old but still heavily used) or uncertain.

This project's approach is based on the 2-refined neutrosophic function developed by Florentin Smarandache, represents each cache item's usefulness using three separate measures instead of one, how useful it clearly is, how uncertain that usefulness is and how contradictory the evidence about it is.

These measures are then tracked over time and evaluated using an improper integral to see whether an item's usefulness settles down or keeps growing. The result of that calculation becomes the basis for the eviction decision forming a new kind of cache replacement policy built on neutrosophic functions and improper integrals.

NEXT