Finding Hierarchical Heavy Hitters in Data Streams

Abstract

Aggregation along hierarchies is a critical summary technique in a large variety of online applications including decision support (OLAP), network management (IP clustering and denial-of-service attack monitoring), text (on prefixes of strings occurring in the text) and XML summarization (on prefixes of root-to-leaf paths in the XML data tree). In these applications, the data is inherently hierarchical and one needs to maintain aggregates at different levels of the hierarchy over time in a dynamic fashion. It formalizes the problem of finding heavy hitters in massive data streams that considers their hierarchical structure. Such hierarchical heavy hitters (HHHs) present a “cartogram” summary of the data stream distribution. This chapter presents comprehensive solutions to the problem of estimating HHHs on data streams. The resulting summary gives a topological “cartogram” of the hierarchical data. It presents deterministic and randomized algorithms for finding HHHs, which builds on existing techniques by incorporating the hierarchy into the algorithms.