An Intelligent agent-based data mining info-structure towards the generation of generic data mining services

Loading...
Thumbnail Image
Date
2004-08
Authors
Hassan, Syed Zahid
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This thesis presents a case for an intelligent agent-based framework for knowledge discovery in a distributed environment comprising multiple heterogeneous data repositories. Data-mediated knowledge discovery, especially from multiple heterogeneous data resources, is a tedious process and imposes significant operational constraints on end-users. We demonstrated that autonomous, reactive and proactive intelligent agents provide an opportunity to generate end user-oriented, packaged, valueadded decision-support/strategic planning services for professionals, managers and policy. makers, without the need for a priori technical knowledge. Since effective services delivery is grounded in good communication, experience sharing, continuous learning and proactive actions, we presented an Agent-based bata Mining Info-structure (ADMI) that deploys a suite of Data Mining (DM) algorithms coupled with Intelligent Agents to facilitate data access, DM query specification, DM algorithm selection and DM result visualisation-i.e. automated generation of data-mediated decision-support/strategicplanning services. We formalized the possible synergy to integrate Agents and Data Mining technologies to generate data-driven strategic services that can go a long way in addressing some of the challenges faced by any modem enterprise. Our methodology purported the exploitation of experiential knowledge, derived from enterprise-wide databases, for strategic decisionmaking. The feasibility of our methodology depends on two factors: (a) the availability of a mass of 'knowledge-rich' data with knowledge of practices and protocols, and (b) the technical capability to extract 'decision-quality' knowledge from data. Firstly, for all practical purposes, modem information retrieval or DM systems generate massive amounts of 'knowledge-rich' data, but unfortunately this asset is not yet fully 'cashed'. Secondly, from a technical point of view, we developed a viable IT info-structure that will help transform data to strategic knowledge services. We also formalized the methodology where the agents in distributed environment can collaborate, cooperate and communicate with each other with the help of one central middle agent. The middle agent provides look-up services (yellow pages) to identify agents with their specific capability description and abetting them to communicate with each other. We also explored and resolved some issues that those information-centric multi-agent place on agent communication languages such as (KQML). The fact that all agents communicate through a common language, allows for an open-ended architecture whereby new DM services and DM algorithms can incrementally be added. In conclusion, we believe that the emerging intelligent agent-based framework provides interesting opportunities to operationalize the volumes of data routinely collected withih. numerous enterprises.
Description
Keywords
Agent-based data mining , Info-structure towards , Generic data mining
Citation