Iknet: Scalable Infrastructure for Achieving an Internetbased Knowledge Network

1
Iknet Scalable Infrastructure for Achieving an
Internet-based Knowledge Network

• Stanley Y. W. Su
• Herman Lam
• Database Systems Research Development Center
• University of Florida
• SSGRR 2000

2
Outline

• Introduction
• Related Research
• Knowledge Model
• Architecture of Knowledge Network
• Concepts of Knowledge Network
• System Implementation
• Summary and Future Research Issues

3
Internet is great, but

• Tedious interactions
• Too many places to go
• Too much data
• Islands of information
• Integrating Web sites are difficult
• Only data, not knowledge
• No know-how, decision-making processes

4
The Knowledge Network Approach

• Add Knowledge into the Internet
• Model knowledge as Events, Triggers, and Rules
• Data providers can provide knowledge
• Data consumers can subscribe to events and also
  install knowledge
• Add replicable components to Web servers

5
Knowledge Network Makes the Internet

• Active
• Automatic response to events, new data data
  operations
• Collaborative
• Spontaneously provide information and initiate
  interactions
• Intelligent
• Embed knowledge into the Internet

6
IKnet Architecture IKhub Components
Browser
App
App
Internet Communication Infrastructure
. . .
IKnet Infrastructure
Overall Iknet architecture
Internet Communication Infrastructure
IKhub
Process Server
DO Server
Other Servers
Rule Server
Web Server
Negotiation Server
Event Server
An IKhub
7
Knowledge Model

• Event-Trigger-Rule (ETR) paradigm
• Event An occurrence of something of interest
• Rule A high-level specification of a granule of
  control and logic
• Trigger Specifies which event structure can
  fire which rule structure
• Advantages
• Events and Rules are independently defined
• Provides flexibility for a distributed environment

8
Event

• Event name ( type)
• Defines the attributes of an event
• Posting modes
• Asynchronous multiple recipients
• Synchronous single recipient
• Sources
• Application-posted, Method-associated, Timer

9
Rule

• Condition-Action-AlternativeAction format
• Rule parameters
• Rule variables
• Temporary, Persistent type, Reference type for
  distributed objects, Customizable type
• Condition Guarded expression
• Action, AlternativeAction Method calls
• Can be enabled/disabled

10
Trigger

• Triggering Event
• OR of Events
• Event History
• Complex event expression to be evaluated against
  event base
• Rule Structure
• Specifies execution order
• Sequential / parallel / AND-OR synchronized

11
Event E3(String make, Integer
discountRate, Date start, Date
end ) Description a promotion for
discounts on car sales
E1
E2
E3
TRIGGER EVENT
E1 or E2 or E3
Rule R3 ( String make,
Integer discount) Description Purchase if
discount is more than 3 for a
Ford RuleVar Integer purchaselimit Condition
makeFord discount gt 3
Action purchaselimitGetLimit()
PurchaseCar(purchaselimit)
AltAction StoreInDB(make,discount)
EVENT HISTORY
E4 gt E5
R1
RULE STRUCTURE
R2
R3
R5
R4
OR1
AND
R6
R7
12
Architecture of Knowledge Network

• Composed of Knowledge Web Servers
• Knowledge Web server
• Knowledge Profile Manager (KPM)
• GUI tool for managing knowledge profiles
  containing definition of events, triggers and
  rules
• Event Manager
• Processes event registration, filtering and
  pushing
• ETR Server
• Processes triggers and executes rules

13
Provider
Event
ETR Server
KPM
Trigger
Knowledge Web Server A
User A
Rule
Filter
Publish Events, Rules
Event Manager
F1
E1, E2, R1, R2
INTERNET
Publish Events, Rules
Publish Events, Rules
E1
Event Manager
Event Manager
KPM
KPM
R3
ETR Server
User B
ETR Server
User C
Knowledge Web Server B
Knowledge Web Server C
Subscriber
Subscriber
14
Concepts of Knowledge Network

• Publishing Knowledge
• Event Filtering and Notification
• Trigger and Rule Processing
• Knowledge Profile Management

15
Publishing Knowledge ...

• Involves
• Defining event
• Defining parameterized filter
• Defining provider-side customizable rules
• Use KPM GUI tools
• Result is an event registration form

16
Publishing Knowledge

• Event Registration Form
• Document-driven form generation approach
• Dynamically convert XML to HTML form
• No coding necessary, just create a link
• Referring to Event Manager
• Input parameters are event filter template and
  provider-side rule template

17
(No Transcript)
18
Event Notification

• Event Notification
• Format E-mail, XML, Java object
• XML format allows interoperability
• Communication HTTP or RMI
• HTTP is already widespread
• Current prototype uses point-to-point
• Straightforward to make use of IP multicasting

19
Trigger and Rule Processing

• Use KPM to define triggers and rules
• When an event is generated at the source
• Event is filtered to identify subscribers
• Provider-side rules are executed
• Event is notified to subscriber site
• Subscriber-side trigger and rules are processed

20
Knowledge Profile Management

• Provider Knowledge Profile
• Published events, parameterized filters,
  customizable rules, simple triggers
• Subscriber Knowledge Profile
• Subscribed events, complex triggers and rules
• Editing Knowledge Profile
• Modify or delete events, triggers, rules
• Delete published event use system-level event
  and rule to clean-up on subscriber site

21
Steps for Constructing the Knowledge Network

• Build-time
• Publishing knowledge by provider
• Event registration by subscriber
• Trigger and rule definition by subscriber
• Run-time
• Event generation and filter processing
• Provider-side rule processing
• Event notification
• Subscriber-side trigger and rule processing

22
Provider Site
Subscriber Site
INTERNET
?
Publish Event/Filter
Event Registration
?
Browser
Browser
?
Publish Rule/Trigger
Rule/Trigger Definition
Knowledge Profile Manager
Knowledge Profile Manager
?
?
Rule /Trigger Definition
?
Event /Filter Definition
Rule /Trigger Definition
Event Registration
Event Mgr
ETR Server
Event Mgr
ETR Server
Web Server
Web Server
Event
?
?
?
Event
Event
Data Applications
Data Applications
?
Event Generation
23
System Implementation ...

• Development Environment
• JDK 1.2.2
• Apache Web server version 1.3.9
• JServ servlet engine version 1.1b3
• Internet Explorer browser version 5.0
• Developed System with Java
• Easily deployable on any Web server on the
  Internet

24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
Summary

• Knowledge Network
• Extend the Internet-Web infrastructure by adding
  event and rule services
• Publishing knowledge
• Event filtering and notification
• Trigger and rule processing
• Knowledge profile management
• Scalability install Multiple knowledge Web
  servers on the Internet

31
Future Research Issues

• Security issues related to event delivery and
  rule processing
• Application domain ontology issues
• Event and rule warehouse
• Efficient event delivery mechanisms