Interoperability among user-adaptive systems in the World Wide Web Francesca Carmagnola [email protected] Reading club 08 May 2007.

Interoperability among user-adaptive systems in the

World Wide Web

Francesca Carmagnola

[email protected]

Reading club 08 May 2007

Master degree in Science and Communication (2004)Communication medium and techniques

CurrentlyPh.D. at Department of Computer Science,   University of Torino, Italy

My research interests – Adaptive Hypermedia Systems– User Modeling– Semantic Web – Web 2.0

What about me?

Member of…

SETA: “Intelligent User Interface” group (http://www.di.unito.it/~seta/). - Design and development of Intelligent Systems, based on distributed architectures, and exploiting Internet technologies-Exploitation of advanced AI techniques to improve the interaction with the users

CIRC (Interdepartmental Centre of Research on Communication) at University of Turin

Social Media Applications Research & Tagging Laboratory (http://www.smartlab.csp.it/)

- Multimedia innovative solutions for mobile users and workers, by developing personalized services

The project which I am involved in….

Web 2.0 + Um

adaptive mobile tourist

guides

DIADIE-tourism

UbiquiTO-S

iCITY

Personalized Digital Television

Interoperability among

user-adaptive systems in

the World Wide Web

(Ph.D. research)

Um + SW

Um + SW

User Modeling + Semantic Web

User Modeling can benefit from techniques provided by SW:

• Standardization of languages

• Reuse of knowledge

• …

Semantic Web can benefit from personalization since it

enables sharing content and services which are tailored to the

needs of individual users (reducing information overload,

etc…)

DIADI

(Project financed by Region Piemonte, Ministry of Economy, EU)

Multi-channel adaptive platform, based on a semantic

representation of contents, for the negotiation

of requests toward providers

•Semantic representation of knowlegde (RDFS)

•Semantic representation of rules (inference rules and adaptation

rules)

•SemanticSearchEngine (used to query the ontologies),

implemented with SeRQL language over a RDF repository

•API delivered http://talea.csp.it/it

UbiquiTO-S

Project Prin/Cofin, financed by Miur, in collaboration with the universities of Udine, Pisa and with the CNR (Trento)

General framework (modular architecture) to develop adaptive mobile tourist guides. In particular, the role of our research group is:

- Study and management of the knowledge (concerning the user, the domain and the context) required in adaptive context-aware mobile guides (semantic modelling)

- Module for the dynamic generation of user interface, on the basis of this knowledge.

Project for the development of a tourist context-aware

guide which provides support and advices to the users

considering the current interaction context.

In particular, the role of our research group is:

- Study of the meaning of the context in tourist context-aware guide (physique, social and personal context)

- Ontological modelling of such a context

E-Tourism

(in collaboration with Telecom Italia lab)

Shift toward Web 2.0

Web 1.0 Web 2.0

Centralized production of contents (top-down approach)

Decentralized production of contents (bottom-up approach)

Peer production

Stand-alone user Community

Social-network

Predefined navigation paths Navigation by tags

Web 2.0 + Adaptation

Web 2.0 Adaptation

Decentralized production of contents (bottom-up approach)

Peer production

Recommendation of contents

Adaptation to the device

Community

Social-network

Recommendation of “similar users” (having similar interests)

Navigation by tags Recommendation of tags

(most personal navigation)

Web 2.0 + Adaptation

Adaptation can benefit from Web 2.0− user participation (insertion of information, tagging, annotations) to learn about the user and thus to create/improve her user model;

Web 2.0 can benefit from adaptation− exploitation of the user model in order to help the user in tagging and creating contents and to support the user in navigation;

− creation of communities of users through the combination of user modeling and participation (especially tagging);

− ….

iCITY

(in collaboration with the Municipality of Torino and SMARTLAB - social media applications research and tagging Laboratory)

• Mobile;

• Adaptive;

• Web-based;

• Social guide;

• Providing personalized information about cultural events of the city of Torino

iCITY - Main functionalities

navigate by categories - by

tags

access cultural events (by text

and map)

Bookmarking preferred events

access personalised

recommendations

Insert content: i)new event, ii)add

information to events, iii) add comments

Tagging event

See and update his user model

See users that inserted content Share

bookmarks

Share tags

Share tags

http://www.icity.di.unito.it/dsa/

Personalized Digital Television(in collaboration with Telecom Italia lab)

Personalization techniques for the customization of the future television services

Customization of the Electronic Program Guides (EPG) focused on the personalized selection of the TV programs to be advertised, on the basis of the user's interests. (Recommendations to help users on orientating into the very large information space available in Digital TV) Use of tagging to refine use model and navigation paths

Exploitation of such techniques within a prototype system for the generation of personalized (EPGs).

Multi-agent architecture to support the development of highly configurable hybrid recommender systems, which integrate different user modelling and recommendation techniques to improve the recommendations to the user.

FINALLY MY PERSONAL RESEARCH…

Interoperability among user-adaptive systems in the

World Wide Web (Ph.D. research)

Starting assumptions

Personalization crucial in many areas (e-learning, tourism, digital libraries, e-commerce, etc.)

The user spends her time interacting with many web-based adaptive systems

Possibility of having common knowledge about the user shared across different systems

Share knowledge about the user across the different systems she interact with!

Advantages of cross-systems personalization

include in the user model features that one system could not acquire by itself (Carmagnola and Cena, 2006)

increase the amount of information about users, since there is the chance to benefit from the efforts led by other modellers and systems. It lets the “increased coverage” since more aspects can be covered by the aggregated user model because of the variety of the contributing systems. Ums themselves can be more accurate and, as a consequence, the adaptation results are improved (quantitative and qualitative improvement) (Stewart, Celik et al. 2006), (Berkovsky 2005)

Advantages of cross-systems personalization

speed up the phase of the user model inizialization (Kobsa, Koenemann et al. 2001)

save the user from tedium of training new systems

reduce the cost of user modelling sharing it among different applications

and many others…

Challenges of cross-systems personalization

Why should commercial competitive systems cooperate?

How to preserve user’s privacy?

How to cope with syntactic and semantic heterogeneity of information over the web?

My research

More specifically…

2. Analysis of the phases required to get user model interoperability

3. Framework providing a common base for the interoperability of user model knowledge among user-adaptive systems in the World Wide Web

1. Analysis of the implications of interoperability in user modeling

Syntactic Semantic

1. Analysis of the implications of interoperability in User Modeling

How to ensure syntactic and semantic interoperability?

Semantic Web techniques

Languages for representing data ( RDF(S), OWL, etc…)

Framework for stucturing data in a syntax-independent way (ontologies)

Languages for reasoning over knowledge (SWRL, OWL-S)

In my framework the User Model knowldege is ontologically represented (RDFS)

2. Analysis of the phases required to get user model interoperability

The cooperation among systems to exchange user model knowledge can be seen as complex task:

A Discovery phase:Finding out the systems which store knowledge on a same user

* The most relevant researches have almost exclusively focused on the core part of the cooperation of user adaptive systems, that is the exchange of data about users

B Exchange (*):Query for the required user data

C Conflict resolution and evaluation:- Need for suitable stategies to cope with semantic interoperability

- Need for suitable stategies to check and solve conflicts and evaluate if the exchanged data are reliable

3. Framework providing a common base for the interoperability of user model knowledge among user-adaptive systems in the World Wide Web

Requirements: every user should be able to declare what information to make

public every system should represent user knowledge in RDF(S) format

and like <p,v> pairs every system should make the repository (containing the

semantic representation of the Um) available to other systems the available information should be updated systems must be able to access, in a simple and efficient way,

to the user information which have been made available by other

systems

Need for a framework for storing, querying and retrieving references to the semantic repositories: SESAME (http://openrdf.org/)

- Open source Java framework that can be used as a database serverwhich client applications can access through the HTTP protocol.

-SeRQL to query repositories-To support Sesame servlet we encode the Java servlet technology in an Apache Tomcat environment.

What already done thus far….

A Discovery phase

Identification algorithm, to discover systems which store knowledge on a same user

2 Java APIs to support designers and systems in performing such a task

B Exchange

Semantic representation of knowledge

Semantic query for the required user data

What I would like to achieve from my visit in Leeds?

C Conflict resolution and evaluation

- Need for suitable strategies to cope with semantic interoperability

(If system A says I like football and system B says I like soccer?)

(If system A says I like football and system B says I like sport?)

When semantic interoperability has been reached….

- Need for suitable strategies to check and solve conflicts and

evaluate if the exchanged data are reliable(If system A says I like football and system B says I don’t like soccer?)(If system A says I like football and volleyball, and system B says I don’t like sports?)(If system A says I like football, and system B says I don’t like football and system C says that I like football?)

ANY SUGGESTION?