Work Package 6

Personalized semantic search

Objectives: to design, develop and evaluate personalized semantic search strategies.

We intend to develop a scalable and generalisable set of search strategies, for example, based on historical events, user interests, spatial navigation preferences and trust policies.

The work in WP06 builds on previous and related work on baseline semantic interoperability between collections and personalized access to cultural heritage collections (e.g. images, audio and video material). The aim is to provide advanced search facilities (e.g. tailored event-based semantic search) in dynamic open linked data (i.e. curated and non-curated Web content), considering content semantics, users domain interests, temporal and spatial preferences and trust and privacy policies. RDF/OWL representations of enriched vocabularies, metadata and digital objects provide a highly interlinked semantic search graph, both in terms of the domain data as well as for the user interest and preference data representation.

The approach will be to deploy “patterns-based” search, where the patterns are semantic-link structures in the graph. The aim is also to provide techniques for clustering and ranking the search results in a for users intuitive fashion. Challenges here are

  1. collecting reliable user data,
  2. building contextualized (time/space) user profiles,
  3. combining the “patterns-based” search with recommendation strategies to meet the users interests and preferences to achieve a personalized presentation of context-based narratives and sequences of objects.

The major role for partners B&G and RMA will be in supporting the evaluation studies.