A Personalized Walk through the Museum: The CHIP Interactive Tour Guide

From ETC Public Wiki
Jump to: navigation, search

A Personalized Walk through the Museum: The CHIP Interactive Tour Guide
Authors: Roes, Stash, Wang, Aroyo
Category: Technology


Design of a system that focused on supporting personalized interactions online and on-site and provide a bridge between the online and on-site experiences for the Rijksmuseum Amsterdam (art museum). Bottleneck for online-onsite bridges had previously been at combining user data from the two spaces.

The project focused on developing an online art recommender which would then be part of online tour wizard which would create a personalized tour of the Rijks. This personal tour could then be loaded onto the Mobile Museum guide onsite.

They focused on optimizing a few aspects:

  • Adapting tour to an allocated time, based on priorities in personalization step
  • Taking real time feedback on which art was enjoyed or not and changed tour to fit these criteria. Optimized for navigation

Resulted to be very effective for novice users to find art in their interest. More studies needed to examine efficiencies.


  • About the system:
    • 3 components: Online Art Recommender, Online Tour Wizard (online demo at: http://chip-project.org/demo) and Mobile Museum Guide (demo: http://chipproject.org/demo/mobileguide).
    • The online Art Recommender (fig. 1 on page 3 of the document) helps users to discover their art interests in the museum collection and to store them in a corresponding user profile.
    • The online Tour Wizard generates online museum tours containing recommended artworks according to the user's interests.
    • The mobile Museum Guide offers the tours created online, provides extensive description of each artwork, as well as a set of related artworks to the ones included in the tour or any given one in the museum. The users can dynamically adapt their tours by expressing in terms of ratings preferences in some seen artworks, or by indicating a desired tour length or number of artworks to be included in the tour.
  • About system architecture:
    • Client-server architecture with Java Servlets running on the CHIP server (described in figure 5 on page 5 of the document). The collection data refers to the enriched museum collections, currently the Rijksmuseum ARIA database, maintained in a Sesame Open RDF memory store and queried with SeRQL. The user data contains OWL user models and tour data. The Web-based demo components are realized as Java Servlets and JSP pages with CSS and JavaScript. The mobile guide is implemented as a Web application running via a Web browser on different mobile devices.
  • About mobile guide:
    • Three main actions for realizing these adaptive steps:
      • Filter out from the tour artworks that do not satisfy the current constraints.
      • Add artworks to the tour that qualify according to contextual and user criteria.
      • Re-order artworks in an optimal navigation sequence (e.g. to fit the room configuration and user position).
    • Collect both implicitly (e.g. monitoring user position and time spent per artwork) and explicitly (e.g. user ratings, indicate the maximum number of artworks the user wants to see in a tour or the maximum amount of time the user wants to spend in the museum).
    • Tour configuration level, the user can:
      • Adapt the number of artworks to see in a tour based on the maximum length of the tour (in minutes) explicitly given by the user.
      • Adapt the time to be spent in a tour based on the maximum number of paintings the user is interested to see in one tour. The user can change those adaptation parameters at any point during the usage of the mobile guide (e.g. before starting a tour, while following a tour).
    • Following the tour: the user indicates interest in a given artwork. In this case, a set of related artworks are offered, and potentially included in the tour. This would result in adding the selected artworks to the tour, re-ordering the current tour to fit the new spatial constraints and filtering out artworks in order to meet temporal or number restrictions given by the user at a configuration level.
    • Finding related artworks using semantic relations between them, e.g. semantic relations between their properties. For examples, related artworks to a self-portrait of Rembrandt would be other portraits and self-portraits and male portraits by Rembrandt, male portraits by students of Rembrandt, male portraits or self-portraits by other artists in the same style.

Back to: Avatar Research