Data Mining Techniques in Grid Computing Environments

4

ADaM services: Scientific datamining in the service-orientedarchitecture paradigm

Rahul Ramachandran, Sara Graves, John Rushing, Ken Keizer, Manil Maskey, Hong Lin and Helen Conover

ABSTRACT

The Algorithm Development and Mining System (ADaM) was originally developed in the early 1990s with the goal of mining large scientific data sets for geophysical phenomena detection and feature extraction. The original design was a comprehensive system that comprised key software components for distributed computing including a mining daemon to handle mining requests, a mining database to fetch and stage appropriate data, a mining scheduler to schedule different mining jobs, a set of mining operations and a mining engine to parse mining plans or workflows and to execute the right mining operations in the right sequence. ADaM has over 100 algorithms and has been used extensively in different applications. With the advent of the service-oriented architecture (SOA) paradigm, ADaM was redesigned and refactored as a toolkit which provides both image processing and pattern recognition algorithms for science data sets as standalone components. These components can be easily packaged as Web or grid services using standard-based protocols. The services can be scripted together to form mining workflows using different composers. The use of standard protocols allows a mining workflow to be deployed as a solutions package at different sites to be reused by others or to be integrated into other applications ...