What can formal concept analysis do for data warehouses?

Abstract

Formal concept analysis (FCA) has been successfully used in several Computer Science fields such as databases, software engineering, and information retrieval, and in many domains like medicine, psychology, linguistics and ecology. In data warehouses, users exploit data hypercubes (i.e., multi-way tables) mainly through online analytical processing (OLAP) techniques to extract useful information from data for decision support purposes. Many topics have attracted researchers in the area of data warehousing: data warehouse design and multidimensional modeling, efficient cube materialization (pre-computation), physical data organization, query optimization and approximation, discovery-driven data exploration as well as cube compression and mining. Recently, there has been an increasing interest to apply or adapt data mining approaches and advanced statistical analysis techniques for extracting knowledge (e.g., outliers, clusters, ules, closed n-sets) from multidimensional data. Such approaches or techniques cover (but are not limited to) FCA, cluster analysis, principal component analysis, log-linear modeling, and non-negative multiway array factorization. Since data cubes are generally large and highly dimensional, and since cells contain consolidated (e.g., mean value), multidimensional and temporal data, such facts lead to challenging research issues in mining data cubes. In this presentation, we will give an overview of related work and show how FCA theory (with possible extensions) can be used to extract valuable and actionable knowledge from data warehouses. © Springer-Verlag Berlin Heidelberg 2009.