knowledge grids akshat mishra grid seminar winter 2008 feb 2008
TRANSCRIPT
KNOWLEDGE GRIDS
• A multinational financial organization has data repositories in various locations allover the world
• The company performs data mining operation such as clustering classification and other knowledge discovery procedures.
• It uses data ming tools for knowlege discovery but faces problem doing co-ordinated work with all its centres.
DISTRIBUTED MINING
• Parallel platforms were used .• Implementing this task on geographically
distributed sites was a challenge• Grid computing could be a possible solution • How could it be done?
Evolution of Data Grids.
• Data Grids store large data sets and are moved with the same ease like files.
• Data Grids evolved into knowledge grid to aid distibuted high performance computation.
• Along with the basic infrastructure of a grid, provided tools for knowledge discovery.
• Knowledge Grids are distributed paralled software architecture supporting knowledge discovery
• Fig
ARCHITECTURE OF KNOWLEDGE GRID
Fig. 1. KNOWLEDGE GRID architectureDistributed Data Mining on Grids:Services Tools and Applications
ARCHITECTURE OF KNOWLEDGE GRID
• Two main services of Core-K Grid layer• Knowledge discovery services• Extends the globus monitoring
service ,manages metadata• The metadata include tools and resources to
be mined.• Mining Tools and algorithms are also mined• Distributed execution plans
ARCHITECTURE OF KNOWLEDGE GRID
• RAEMS(Resoe allocation and exection management services)
• To find suitable mapping between execution plan and available resources
• The goal is to satisfy application requirements• The layer can also use the facility of GRAM
ARCHITECTURE OF KNOWLEDGE GRID
• HIGHER LEVEL K GRID LAYER• DAS(Data Access Service)• DAS is reponsible for
searching ,selecting ,extracting,transforming and deleivering data to be mined
• It is based on user requirement• TAAS(Tools and algorithms access service)• TAAS is responsible for downloading algos &
tools
ARCHITECTURE OF KNOWLEDGE GRID
• EPMS(Execution plan management Service)• An execution plan is represented by graph• Graph describes data flow.• It is performed by a semi automatic tool• The tool takes data and programs selected by
the user and generates an abstract exectution plan
• Finally RPS analyzes the result pattern
Design proces of Data Mining
• fig2
Fig. 2. Design process of a data mining computationDistributed Data Mining on Grids: Services Tools and Applications
VEGA SOFTWARE MODULES
Vega Software ModulesDistributed Data Mining on Grids: Services Tools and
Applications
GRAPHICAL REPRESENTATION
• Visual Interface of VegaDistributed Data Mining on Grids: Services Tools and Applications
FIGURE 3 :Dis
Another defintion of Knowledge Grid
• According to H.Zhuge knowledge grid is an intelligent sustainable internet application that enables people or virtual roles to effectively capture ,publish ,share and manage explicit knowledge resources
• It also provides on demand services to support innovation cooperative teamwork ,problem solving and decision making.
WHY KNOWLEDGE GRID
•Evolving of Internet meant that the number of pages kept on increasing•Info in the form blogs, forums etc are exploding on the net•The large explode of informaiton meant there is need to have a information service that spans over an administrative domain.
Evolving concept
• An emerging technology and evolving concept• Google ---->SEARCH ENGINE• YAHOO----->INDEXING INFORMATION ON THE
BASIS OF CATEGORY• Semantic web --->provides services along with
information services
References
• Distributed Data Mining on Grids:Services Tools and applicatios ,
• IEEE TRANSACTIONS ON Systems,MAN and Cybernetics,Vol 34,No6,December 2004
• The knowledge grid,By Mario Cannataro and Domenico Talia
• http://en.wikipedia.org/wiki/Knowledge_Grid• A knowledge grid model and and platform for
global knowledge sharing By H.Zughe• Grid Portals By Fugang Wang• http://cyberaide.googlecode.com/svn/trunk/lectur
es/GridPortal.ppt