Automatic transformation of multi-dimensional web tables into data cubes

  • Authors:
  • Norah Alrayes;Wo-Shun Luk

  • Affiliations:
  • School of Computing Science, Simon Fraser University, BC, Canada;School of Computing Science, Simon Fraser University, BC, Canada

  • Venue:
  • DaWaK'12 Proceedings of the 14th international conference on Data Warehousing and Knowledge Discovery
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

The similarities between data cubes and multi-dimensional tables have long been noted. Routinely, OLAP reporting tools produce multidimensional tables from data cubes. In this paper, we develop a scheme that does the reverse transformation, automatically, so that one may produce charts directly from multi-dimensional tables using standard OLAP data visualization tools. In the process, we develop several new techniques for table processing: (i) extraction of non-overlapping hierarchies from a table; (ii) extraction of metadata from the table title via natural language processing; and (iii) integration of tables in a table series, and integration of tables with common dimensions. Experiments were conducted on some 800 summary tables from Statistics Canada, and our success rate was greater than 90 tested.