Diapositiva 1

1
EADTU Seminar Innovations in supporting distance
learners in Europe Towards personalisation in
digital libraries through ontologies
9th May 2005 ,Open University Library Learning
Resources Centre Núria Ferran Julià
Minguillón, Universitat Oberta de Catalunya (UOC)
2
Main Goal of the Project
The basic idea is that efforts for finding a
useful piece of information in a Digital Library
carried out by an individual can be stored in
a structured way and then shared for future
users with similar necessities.

• The main goal is to help the users of a digital
  library to improve their
• experience of use with personalised services by
  means of two
• complementary strategies
• by maintaining a complete history record of his
  or her browsing and searching activities, which
  is part of a user profile,
• by reusing all the knowledge which has been
  extracted from previous usage from other library
  users with similar profiles.
• Then, all these profiles are combined with the
  help of an ontology.
• All this can be accomplished through the use of a
  recommendation
• system.

3
Personalisation

• Personalisation is one of the key factors which
  are directly related to user satisfaction.
• Personalisation has been shown useful in several
  areas.
• In an ideal scenario, the DL should adapt to the
  specific characteristics of each user profile,
  but also to the particular necessities and
  preferences of each user, combining both user and
  profile level personalisation capabilities.
• In order to build the personalisation system
• Cognitive and behavioral aspects that determine
  the way users perform searches and examine the
  obtained results
• Must be addressed from a user-centered approach
  (HCI)
• Technological and knowledge engineering aspects
  related to the way all this information is
  structured for both updating and querying.
• This project describes the set of desidered
  functionalities and requirements
• of an ideal scenario for a DL which includes
  personalisation capabilities by
• means of ontologies.

4
Ontology

• Is a formal description of a possible scenario or
  context.
• We use the ontology not for describing the
  contents of a library but for describing the way
  users browse and search such contents, with the
  aim of building a personalisation system based on
  accurate recommendations.
• Collaborative filtering selects content based on
  user preferences by polling and ranking informed
  opinions (or experiences of use) on any topic.
• Ontologies and taxonomies are often used as
  synonyms. Ontologies include a set of semantic
  rules which are used to infer knowledge from a
  structured hierarchy of information, giving to
  the complete structure a semantic meaning, not
  only syntactic.
• Ontologies are built using other sub-ontologies
  which describe the basic elements of the
  personalisation system
• users,
• digital resources,
• actions,
• etc.

5
User profile

• Should include all the information relevant to
  user
• Personal information (which can be public)
• Navigational history and behavior records
• Preferences and community involment.
• This information should help to improve the
  searchers by obtaining
• additional information from the recommender
  system.
• This information has been validated by the
  ontology and that is not
• biased by any non-academic purpose of use (such
  as commercial
• supported recommendations in Google or Amazon).

6
Basic user profile attributes for building the
user model
7
Information sources and basic user actions in the
digital library
The actions that a user perform are the implicit
way of the recommender system for determining the
importance of each content.
Therefore, recommendations are generated using
the knowledge extracted from the searching and
browsing profiles of users with similar
interests.
8
Privacy issues

• The users are always under control in a virtual
  e-learning
• environment, all actions are monitored and
  registered.
• users know in advance that all actions are
  logged
• the recommendation system must be designed in a
  non-intrusive manner and be user-friendly,
  including the possibility of disconnecting it or
  minimizing its participation in the navigation
  activities
• the participation of each individual user in the
  final recommendation system is completely
  anonymous
• the collected information is only used with
  personalization purposes.

The more information the user reveals, the more
personalised services he or she obtains.
9
Basic steps for designing an ontology
10
Qualitative usage information

• The digital resources which are catalogued
  through standards can be extended by means of an
  ontology to include additional information, such
  as their usage. Through the collaborative
  filtering all these information about the use of
  each information resource can be stored as
  qualitative information.
• When users navigate among internal sources the
  qualitative information can be easily
  incorporated into the navigational and
  recommender system.
• When users navigate among external sources, there
  is a need establishing an interchange between
  information providers and the DL in order to keep
  the qualitative usage information.
• The DL stores
• the User ID
• Time Session
• The Information Provider sends
• Resource Identifier (URI, URL, DOI, ISBN, etc.)
• Navigational Action Identifier (i. e. searching,
  downloading, asking a loan).

With this proposal we can rate the quality of
each information resource through the history of
navigation actions performed by different user
profiles. This can only happen if the DL is
used as the starting point for navigation!
11
Conclusions

• Ontologies are a powerful tool for describing
  complex scenarios of use such as a DL. The use of
  ontologies promotes the integration of new
  services into existing ones, and the
  interoperability with other systems through the
  appropriate Semantic Web services.

• New system functionalities and requirements can
  be added by including the appropriate description
  into the ontology framework that defines the
  digital library scenario of use.

• Current and further research in this subject
  include the integration of the DL personalisation
  services with other personalisation mechanisms
  provided by the virtual campus, towards a unique
  and complete user model.

• The definition of a validation rating algorithm
  combining both automatic but also user explicit
  rating systems is also under consideration.

12
Further information Núria Ferran nferranf_at_uoc.
edu