Memory Efficient Algorithm for Mining Recent Frequent Items in a Stream

In the paper we present an improved version of multistage hashing based algorithm, used to find frequent items in a stream. Our algorithm uses low-pass filters instead of simple counters, so it concentrates more on recent items and ignores the old ones. Such behaviour is similar to sliding window based algorithms, but requires less memory and is suitable for real-time applications. The algorithm continuously gives estimates of frequencies of the most frequent items. It was tested with streams having various frequency distributions and proved to work