Big Data

Modularizing Data Mining: A Case Study Framework

Date Added: Jul 2009
Format: PDF

This paper presents the fundamental concepts underpinning MoLS, a framework for exploring and applying many variations of algorithms for one datamining problem: mining a database relation for Approximate Functional Dependencies (AFDs). An engineering approach to AFD mining suggests a framework which can be customized with plug-ins, yielding targetability and improved performance. This paper organizes familiar approaches for navigating a search spaces and introduces a new concepts to define and utilize variations of those spaces.