Intelligent Web Applications

Intelligent Web Applications (Part 1) Course Introduction

Vrije Universiteit Amsterdam, Fall 2002 Vagan Terziyan AI Department, Kharkov National University of Radioelectronics / MIT Department, University of Jyvaskyla [email protected] ; [email protected] http://www.cs.jyu.fi/ai/vagan/index.html +358 14 260-4618

Contents ❧ Course Introduction ❧ Lectures and Links ❧ Course Assignment ❧ Examples of course-related research

2

Course (Part 1) Formula: Web Personalization + Web Mining + + Semantic Web + Intelligent Agents = = Intelligent Web Applications - Why ? - To be able to intelligently utilise huge, rich and shared web resources and services taking into account heterogeneity of sources, user preferences and mobility. - What included ? - Introduction to Web content management. Web content personalization. Filtering Web content. Data and Web mining methods. Multidatabase mining. Metamodels for knowledge management. E-services and their management in wired and wireless Internet. Intelligent e-commerce applications and mobility of users. Information integration of heterogeneous resources. 3

Practical Information ❧ 9 Lectures (2 x 45 minutes each, in English) during period 28 October - 15 November according to the schedule; ❧ Course slides: available online plus hardcopies; ❧ Practical Assignment (make PowerPoint presentation based on a research paper and send electronically to the lecturer until 10 December); ❧ Exam - there will be no exam. Evaluation mark for this part of the course will be given based on the Practical Assignment 4

Introduction: Semantic Web - new Possibilities for Intelligent web Applications

5

Motivation for Semantic Web Web Limitations Average WWW searches examine only about 25% of potentially relevant sites and return a lot of unwanted information

Semantic Web Doubles in size every six months

The Semantic Web is a vision: the idea of having data on the Web defined and linked in a way that it can be used by machines not just for display purposes, but for automation, integration and reuse of data across various applications.

World Wide Web

Information on web is not suitable for software agents 4

Semantic Web Structure

Before Semantic Web

Semantic Annotations

Ontologies

Logical Support

Languages

Tools

Applications / Services

Semantic Web

WWW and Beyond

Creators

Users

WWW and Beyond

Web content 7

Creators

Users Web content

6 8

Semantic Web Content: New “Users” Semantic Web and Beyond

Users

Creators Semantic Web content

applications agents

Semantic Annotations

Ontologies

Logical Support

Languages

Tools

Applications / Services

Semantic Web

WWW and Beyond

Creators

Users Web content 7

Some Professions around Semantic Web AI Professionals

Content creators

Content Mobile Computing Professionals

Logic, Proof and Trust

Web designers

Ontologies Agents

Annotations

Ontology engineers Software engineers

8

Semantic Web: Resource Integration

Semantic annotation Shared ontology

Web resources / services / DBs / etc.

9

What else Can be Annotated for Semantic Web ?

External world resources

Web resources / services / DBs / etc.

Web users (profiles, preferences)

Shared ontology

Web agents / applications Web access devices 10

Word-Wide Correlated Activities Semantic Web Semantic Web is an extension of the current web in which information is given well-defined meaning, better enabling computers and people to work in cooperation

Agentcities is a global, collaborative effort to construct an open network of on-line systems hosting diverse agent based services.

Agentcities Grid Computing Wide-area distributed computing, or "grid” technologies, provide the foundation to a number of large-scale efforts utilizing the global Internet to build distributed computing and communications infrastructures.

Web Services WWW is more and more used for application to application communication. The programmatic interfaces made available are referred to as Web services. The goal of the Web Services Activity is to develop a set of technologies in order to bring Web services to their full potential

FIPA FIPA is a non-profit organisation aimed at producing standards for the interoperation of heterogeneous software agents.

11

University of Jyvaskyla Experience: Examples of Related Courses

Distributed Artificial Intelligence in Mobile Environment (2 ov.) Lecturer: Vagan Terziyan

Intelligent Information Integration in Mobile Environment (4 ov.)

Intelligent Web Applications (2 ov.)

Lecturer: Vagan Terziyan University of Jyvaskyla, MIT Department, Spring 2002

Lecturer: Vagan Terziyan

University of Jyvaskyla, MIT Department, Fall 2001, 2002 Vrije Universiteit Amsterdam, AI Department, Fall 2001

Web Content Management (6 ov.)

Vrije Universiteit Amsterdam, AI Department, Fall 2002

Digitaalisen median erityiskysymyksiä (2 ov) seminaarin aihepiiri:

Lecturer: Vagan Terziyan

Semanttinen web

Jyvaskyla Polytechnic, Spring 2002

Lecturer: Airi Salminen

Structured Electronic Documentation Lecturer: Matthieu Weber

University of Jyvaskyla, CS & IS Department, Spring 2002 18

University of Jyvaskyla, MIT Department, Fall 2001, 2002 [email protected] 18

IWA Course (Part 1): Lectures

13

Lecture 1: Web Content Personalization Overview Personalizing Web Resources for a User one of the basic abilities of an intelligent agent

Web Content Personalization Overview

Users Web Resource

3

Based on the Tutorials of K. Garvie Brown, R. Wilson, M. Shamos and others

http://www.cs.jyu.fi/ai/vagan/Personalization.ppt 14

1

Lecture 2: Collaborative Filtering Improving Personalized Service based on Feedback from Users (Collaborative Filtering)

- one of the basic abilities of an intelligent agent 2. Feedback

Web Resource

Collaborative Filtering

Users

1. Recommendation 3. Better recommendations

...

Partially based on tutorials and approaches of GroupLens, Mginetechnologies and Web Museum research groups

http://www.cs.jyu.fi/ai/vagan/Collaborative_Filtering.ppt 15

Lecture 3: Dynamic Integration of Virtual Predictors Discovering Knowledge from Data - one of the basic abilities of an intelligent agent

Data

Dynamic Integration of Virtual Predictors

Knowledge

Vagan Terziyan University of Jyvaskyla, Finland 2

e-mail: [email protected] http://www.cs.jyu.fi/ai/vagan/index.html

http://www.cs.jyu.fi/ai/vagan/Virtual_Predictors.ppt 16

Lecture 4: Introduction to Bayesian Networks Discovering Casual Relationship from the Dynamic Environmental Data and Managing Uncertainty - are among the basic abilities of an intelligent agent

beliefs

Introduction to Bayesian Networks

Casual network with Uncertainty

Dynamic Environment Based on the Tutorials and Presentations:

2

(1) Dennis M. Buede Joseph A. Tatman, Tatman, Terry A. Bresnick; Bresnick; (2) Jack Breese and Daphne Koller; Koller; (3) Scott Davies and Andrew Moore; (4) Thomas Richardson (5) Roldano Cattoni (6) Irina Rich

http://www.cs.jyu.fi/ai/vagan/Bayes_Nets.ppt 17

Lecture 5: Web Mining Discovering Knowledge from and about WWW is one of the basic abilities of an intelligent agent WWW Knowledge

Web Mining Based on tutorials and presentations: 2

J. Han, D. Jing, W. Yan, Z. Xuan, M. Morzy, M. Chen, M. Brobbey, N. Somasetty, N. Niu, P. Sundaram, S. Sajja, S. Thota, H. Ahonen-Myka, R. Cooley, B. Mobasher, J. Srivastava

http://www.cs.jyu.fi/ai/vagan/Web_Mining.ppt 18

Lecture 6: Multidatabase Mining Discovering Knowledge from Distributed and Heterogeneous Databases - is one of the basic

DB 1

abilities of an intelligent agent Distributed and heterogeneous databases

x

•••

DB n

Classifier 1 •••

?

Classifier m

Knowledge

Multidatabase Mining Based on tutorials and presentations: J. Han, C. Isik, M. Kamber, A. Logvinovskiy, S. Puuronen, V. Terziyan

http://www.cs.jyu.fi/ai/vagan/MDB_Mining.ppt 19

Lecture 7: Metamodels for Managing Knowledge Metamodels for Managing Knowledge

Creating and Managing Knowledge According to Different Levels of Possible Context - are among the basic abilities of an intelligent agent Metacontexts

Contexts

Meta-metaknowledge

Metaknowledge

Vagan Terziyan Data

University of Jyvaskyla, Finland

Knowledge

2

e-mail: [email protected] http://www.cs.jyu.fi/ai/vagan/index.html

1

http://www.cs.jyu.fi/ai/vagan/Metamodels.ppt 20

Lecture 8: Knowledge Management Making Personal Knowledge Available to Others and Dealing with Knowledge Taken from Multiple Sources

- are among the basic abilities of an Intelligent Agent

Knowledge Management Based on tutorials and presentations: R. Bergmann, M.M. Richter, D.J. Skyrme, Bellanet Int’l, SURF-AS, R. L. Herting, R. Smith, F.J. Kurfess, R. Dieng at al., M. Sintek, A. Abecker, A. Bernardi, D. Karagiannis, R. Telesko, L. Kerschberg “Give a man a fish - feed him for a day; teach him how to fish - feed him for a lifetime” Chinese proverb

http://www.cs.jyu.fi/ai/vagan/Knowledge_Management.ppt 21

Lecture 9: E-Services in Semantic Web Managing Transactions with Distributed E-Services and providing Integrated Service to a User - are among the basic abilities of an Intelligent Agent

E-Services in Semantic Web Vagan Terziyan MIT Department, University of Jyvaskyla / / AI Department, Kharkov National University of Radioelectronics [email protected] http://www.cs.jyu.fi/ai/vagan +358 14 260-2347

http://www.cs.jyu.fi/ai/vagan/E-Services.ppt 22

IWA Course (Part 1): Practical Assignment

23

Practical assignment in brief ❧ Students are expected to select one of below recommended papers, which is not already selected by some other student, register his/her choice from the Course Assistant and make PowerPoint presentation based on that paper. The presentation should provide evidence that a student has got the main ideas of the paper, is able to provide his personal additional conclusions and critics to the approaches used. 24

Evaluation criteria for practical assignment ❧ Content and Completeness; ❧ Clearness and Simplicity; ❧ Discovered Connections to IWA Course Material; ❧ Originality, Personal Conclusions and Critics; ❧ Design Quality. 25

Format, Submission and Deadlines ❧ Format: PowerPoint ppt. (winzip encoding allowed), name of file is student’s family name; ❧ Presentation should contain all references to the materials used, including the original paper; ❧ Deadline - 10 December 2002; ❧ Files with presentations should be sent by e-mail to Vagan Terziyan ([email protected] AND [email protected]); ❧ Notification of evaluation - until 15 December. 26

Papers for Practical Assignment (1) ❧ Paper 1: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_1_P.pdf ❧ Paper 2: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_2_P.pdf ❧ Paper 3: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_3_CF.ps ❧ Paper 4: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_4_CF.pdf ❧ Paper 5: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_5_MW.pdf ❧ Paper 6: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_6_BN.ps ❧ Paper 7: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_7_BN.pdf ❧ Paper 8: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_8_MM.pdf 27

Papers for Practical Assignment (2) ❧ Paper 9: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_9_WM.ps ❧ Paper 10: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_10_WM.pdf ❧ Paper 11: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_11_III.pdf ❧ Paper 12: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_12_III.pdf ❧ Paper 13: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_13_KM.pdf ❧ Paper 14: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_14_ES.pdf ❧ Paper 15: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_15_MDB.pdf ❧ Paper 16: http://www.cs.jyu.fi/ai/vagan/course_papers/Paper_16_MDB.pdf 28

University of Jyvaskyla Experience: Examples of Course-Related Research

29

Mobile Location-Based Service in Semantic Web M-Commerce LBS system

Multimeetmobile Project (2000-2001) Academy of Finland Project (1999):

Information Technology Research Institute (University of Jyvaskyla):

Dynamic Integration of Classification Algorithms

Customer-oriented research and development in Information Technology http://www.titu.jyu.fi/eindex.html

Multimeetmobile (MMM) Project (2000-2001): Location-Based Service System and Transaction Management in Mobile Electronic Commerce http://www.cs.jyu.fi/~mmm

18

Adaptive interface for MLS client

http://www.cs.jyu.fi/~mmm In the framework of the Multi Meet Mobile (MMM) project at the University of Jyväskylä, a LBS pilot system, MMM Location-based Service system (MLS), has been developed. MLS is a general LBS system for mobile users, offering map and navigation across multiple geographically distributed services accompanied with access to location-based information through the map on terminal’s screen. MLS is based on Java, XML and uses dynamic selection of services for customers based on their profile and location. Virrantaus K., Veijalainen J., Markkula J., Katasonov A., Garmash A., Tirri H., Terziyan V., Developing GIS-Supported Location-Based Services, In: Proceedings of WGIS 2001 - First International Workshop on Web Geographical Information Systems, 3-6 December, 2001, Kyoto, Japan, pp. 423-432. 19

Only predicted services, for the customer with known profile and location, will be delivered from MLS and displayed at the mobile terminal screen as clickable “points of interest” 20

Inductive learning of customer preferences with integration of predictors

Route-based personalization

< xt1, xt 2 ,..., xtm > Sample Instances

Learning Environment Predictors/Classifiers

< xr1, xr 2 ,..., xrm → yr >

P1

P2

...

Pn

yt Terziyan V., Dynamic Integration of Virtual Predictors, In: L.I. Kuncheva, F.

Static Perspective

Dynamic Perspective

21

Steimann, C. Haefke, M. Aladjem, V. Novak (Eds), Proceedings of the International ICSC Congress on Computational Intelligence: Methods and Applications - CIMA'2001, Bangor, Wales, UK, June 19 - 22, 2001, ICSC Academic Press, Canada/The Netherlands, pp. 463-469.

30 22

Mobile Transactions Management in Semantic Web Web Resource/Service Integration:

Web Resource/Service Integration: Server-Based Transaction Monitor

Mobile Client-Base Transaction Monitor Web resource / service

Web resource / service

Server

Client

TM wireless Client

Server

wireless

wireless

TM

Web resource / service

Web resource / service

Transaction Service

Server

The conceptual scheme of the ontology-based transaction management with multiple eservices

Web Resource/Service Integration: Comparison of Architectures ❧ Server-based TM ●

Positive:

❧ Less wireless (sub)transactions ❧ Rich ontological support ❧ Smaller crash, disconnection vulnerability ●

Negative:

❧ Pure customer’s trust ❧ Lack of customer’s awareness and control ❧ Problematic TM’s adaptation to the customer

❧ Client-based TM ●

Positive:

❧ Customer’s firm trust ❧ Customer’s awareness and involvement ❧ Better TM’s adaptation to the customer ●

Server

20

Client 1 Transaction data

Client r Services data

Transaction data

Negative:

❧ More wireless (sub)transactions ❧ Restricted ontological support ❧ High crash, disconnection vulnerability 22

Services data

Parameter 1

Recent value

Service 1 ********

Parameter 1

Recent value

Service 1 ********

Parameter 2

Recent value

Service 2 ********

Parameter 2

Recent value

Service 2 ********

… Parameter n

…

…

Recent value

…

Service s ********

Transaction monitor

…

Parameter n

…

…

Recent value

Service s ********

Transaction monitor

Ontologies Service atomic action ontologies

Parameter ontologies Parameter 1

Name 1

Default value / schema 1

Parameter 2

Name 2

Default value / schema 2

…

…

…

Parameter n

Name n

Default value / schema n

input parameters

input parameters

Name of action 1

Name of action 2

Service 1

output parameters

output parameters

Service s

Subtransaction monitor

Service Tree

input parameters

Name of action k …

output parameters

Terziyan V., Ontology-Driven Transaction Monitor for Mobile Services, In: Proceedings of Semweb@KR2002 Workshop on Formal Ontology, Knowledge Representation and Intelligent Systems for the World Wide Web, Toulouse, France, 19-20 April, 2002.

21

Subtransaction monitor

Clients data

Service Tree

Clients data

Client 1 ********

Client 1 ********

Client 2 ********

Client 2 ********

… Client r ********

…

…

Client r ********

23

31

P-Commerce in Semantic Web Clients Public merchants, public customers, public information providers

…

I

I

C

S

I MetaProfiles

…

External Environment

Server

XML WML

I

Maps

Maps

<path network>



SMOs SMRs

… Map Content Providers Server

Integration, Analysis, Learning

Business knowledge

Profiles

Negotiation, Contracting, Billing

Location Providers Server XML

… Content Providers Server …

$ $ $ Banks

Terziyan V., Architecture for Mobile P-Commerce: Multilevel Profiling Framework, IJCAI-2001 International Workshop on "E-Business and the Intelligent Web", Seattle, USA, 5 August 2001, 12 pp.

32

Semantic Metanetwork for Metadata Management A''

2

L''

L''

1

2

A''3

A''1 A' 1

A' L' 2

L' 1 A'

3

A'

L' 3

2

L1 A

1

A

L

2

L

L

2

A 3

3

4

4

Semantic Metanetwork is  considered formally as the  Second level set of semantic networks,  which are put on each other  in such a way that links of  every previous semantic  First level network are in the same  time nodes of the next  network. In a Semantic Metanetwork  every higher level controls  Zero level semantic structure of the  lower level.

Terziyan V., Puuronen S., Reasoning with Multilevel Contexts in Semantic Metanetworks, In: P. Bonzon, M. Cavalcanti, R. Nossun (Eds.), Formal Aspects in Context, Kluwer Academic Publishers, 2000, pp. 107-126.

33

Petri Metanetwork for Management Dynamics P´2

t´1 P´1

P´3

t´3 P´4

Controlling level

P´5

t´2

t1

Basic level

P1 P2

• Each level of the new structure  is an ordinary petrinet of some  traditional type. • A basic level petrinet  simulates the process of some  application.

P4

P3

•A metapetrinet is able not only  to change the marking of a  petrinet but also to reconfigure  dynamically its structure

t2

Terziyan V., Savolainen V., Metapetrinets for Controlling Complex and Dynamic Processes, International Journal of Information and Management Sciences, V. 10, No. 1, March 1999, pp.13-32.

• The second level, i.e. the  metapetrinet, is used to simulate  and help controlling the  configuration change at the  basic level.

34

Bayesian Metanetwork for Management Uncertainty Two-level Bayesian Metanetwork for managing conditional dependencies

Two-level Bayesian Metanetwork for managing conditional dependencies

Contextual level A

X

Q B

Y

S

R

Q B

Y

S

2-level Bayesian Metanetwork for modelling relevant features’ selection

X

A

Predictive level

R

3-level Bayesian Metanetwork for Managing Feature Relevance

Contextual level A

X

Q B

Predictive level

S

Y

R

X

A Q B

S

Y

R

Terziyan V., Vitko O., Bayesian Metanetworks for Mobile Web Content Personalization, In: Proceedings of 2nd WSEAS International Conference on Automation and Integration (ICAI’02), Puerto De La Cruz, Tenerife, December 2002.

35

Multidatabase Mining based on Metadata ONE:MANY DB

MANY:MANY

Classifier 1

ONE:ONE

•••

DB 1 •••

Classifier m DB

DB n

Classifier 1 •••

Classifier m

MANY:ONE

Classifier

DB 1 •••

DB n

Classifier

Puuronen S., Terziyan V., Logvinovsky A., Mining Several Data Bases with an Ensemble of Classifiers, In: T. BenchCapon, G. Soda and M. Tjoa (Eds.), Database and Expert Systems Applications, Lecture Notes in Computer Science, Springer-Verlag, V. 1677,

36