Dynamic Aggregation to Support Pattern Discovery: A case study with web logs

Abstract

Rapid growth of digital data collections is overwhelming the capabilities of humans to comprehend them without aid. The extraction of useful data from large raw data sets is something that humans do poorly because of the overwhelming amount of information. Aggregation is a technique that extracts important aspect from groups of data thus reducing the amount that the user has to deal with at one time, thereby enabling them to discover patterns, outliers, gaps, and clusters. Previous mechanisms for interactive exploration with aggregated data was either too complex to use or too limited in scope. This paper proposes a new technique for dynamic aggregation that can combine with dynamic queries to support most of the tasks involved in data manipulation. (UMIACS-TR-2002-26) (HCIL-TR-2001-27)