An open multi-tier architecture for high-performance data mining using SOA

Abstract

Building scalable, extensible, interoperable, distributed and easy-to-use large-scale data mining applications has proved to be challenging. Service-oriented architecture (SOA) is a flexible set of design principles used for solving such challenges. Many studies try to leverage it in building data mining applications. Our paper aims to propose an open multi-tier software architecture using SOA open standards in a homogeneous and distributed environment that goes beyond traditional n-tier software architecture efforts. This architecture uses enterprise service bus (ESB) model by utilising Windows Communication Foundation (WCF) to offer a number of endpoints for published services. All the tiers are independent to convey its functionality through this ESB. As a result, in the context of data mining, this framework relies on web services to achieve extensibility and interoperability, offers simple abstractions for users, and supports computationally intensive processing on large amounts of data.