3. Research Questions
□ Can we use semantic
metadata of cultural
heritage to improve
personalized access
through multiple
devices (e.g. PC,
mobile phone, PDA)?
4. I: CHIP Approach to Semantics
□ Making museum metadata available in RDF/OWL
□ Making relevant vocabularies available in
RDF/OWL
□ Aligning & enriching vocabularies/metadata
□ Using resulting RDF/OWL graph for building a
combined (virtual and physical) user model
□ Using the above result for (semi)automatic
generation of virtual and physical museum tours
6. style: Baroque
teacher of: Gerrit Dou
teacher of: Nicolaes Maesteacher of: Ferdinand Bol
self-portrait
militia
place:
Amsterdam
period:
1625 to 1650
7. RDF/OWL Graph for Building
User Model
□ Users rating:
■ Artworks & artwork properties
■ 20 pre-selected by museum curators
■ rest random
□ Ex.:
■ I like “Jewish Bride”
■ I like creator “Rembrandt”
■ I like style “Baroque”
8. User Profile
□ Based on Web
standards (FOAF,
RDF)
□ Combining physical
& digital activities
from multiple
sources
□ Reusable
□ Distributed
10. Serendipity in Recommendations
□ Recommendations based on explicit semantics (for
reusability)
□ Cold-start problem improved
□ User control in what to consider for recommendations
□ Non-trivial related recommendations
□ Faceted browsing
□ Combination of CB and collaborative aspects
14. Using RDF/OWL User Model for
Generation of Museum Tours
□ Generation of Tours, e.g.
“Masterpieces”, “Favorites”, etc.:
□ positively rated creators
□ related artists, i.e. collaborator & student_of
□ same style & style siblings
□ …
□ topical tours
□ length settings
□ spatial adaptation
17. III: Domain & User-centered
Approach
□ Four experiments:
■ I: Recommendations effectiveness: novices vs. experts; AR
helps novices to elicit/clarify preferences from their implicit
knowledge of the collection.
■ II: Rating, sparsity and cold-start; "expert-sorted" sequence of
artworks works very well for first-time users to build user profiles;
"rating artwork and topics" increases the total amount of the user's
contributions and improves precision of recommendations.
■ III: Accuracy and interestingness of recommendations;
semantic relations can enhance the CB recommendations.
■ IV: Ontology-enhanced content-based & collaborative
recommendations; semantic relations do not give a significant
difference for the performance of recommender.
□ Study: interactive museum tours in NL and online guides
□ Study: in-door navigation & identification techniques
□ Planned experiments: tour adaptation with semantic & spatial
restrictions
18. Research Topics
□ Recommendations with respect to subject/theme
□ Recommendations with respect to semantic relations
□ Recommendations for groups of users
□ Taking into account temporal preferences in
recommendations, e.g. “I am not interested NOW”
□ Adapting the tour sequence on the fly
■ based on position, time, interest, previous activities
□ Adapting the artwork sequence in user profiler
■ improving the cold-start problem
□ Social aspects based on FOAF profile
□ Extension of user profile FOAF specification with
interest property
20. The Team
□ Lora Aroyo
□ Paul De Bra
□ Peter Gorgels
□ Cathy Jager
□ Lloyd Rutledge
□ Mettina Veenstra
□ Natalia Stash
□ Yiwen Wang
21. Follow-up (1)
□ CATCH2 Agora project
■ social platform with CH objects in explicit
(art)historic context
□ digitally mediated public history
□ historic events model for semantically
relevant relationships between objects
□ community curated historic event
thesaurus linked to museum artifacts
■ Rijksmuseum Amsterdam & Dutch
audiovisual archive Beeld en Geluid
22. Follow-up (2)
□ CATCHPlus Rijksmuseum deployment
■ Q42, Fabrique
■ CHIP API
■ CATCH Central User Profiling Service
■ Rijksmuseum “Genius” on
Rijksmuseum.nl