User-cognizant multidimensional analysis

Our goal is to enhance multidimensional database systems with a suite of ad vanced operators to automate data analysis tasks that are currently handled through manual exploration. In this paper, we present a key component of o ur system that characterizes the information content of a cell based on a u ser's prior familiarity with the cube and provides a context-sensitive expl oration of the cube. There are three main modules of this component. A Trac ker, that continuously tracks the parts of the cube that a user has visited . A Modeler, that pieces together the information in the visited parts to m odel the user's expected values in the unvisited parts. An Informer, that p rocesses user's queries about the most informative unvisited parts of the c ube. The mathematical basis for the expected value modeling is provided by the classical maximum entropy principle. Accordingly, the expected values a re computed so as to agree with every value that is already visited while r educing assumptions about unvisited values to the minimum by maximizing the ir entropy. The most informative values are defined as those that bring the new expected values closest to the actual values. We believe and prove thr ough experiments that such a user-in-the-loop exploration will enable much faster assimilation of all significant information in the data compared to existing manual explorations.