Core Research Question
How can attention data be exploited for improved information delivery services?
- Widespread desktop search engines neglect local retrieval knowledge
- Relevancy is determined by standard similarity measures.
- Personalization is a weakly-integrated add-on (query expansion, presentation tailoring).
- Deep integration of context knowledge adapts information retrieval to the user’s needs and views
- Informed local search realizes user-perspective similarity measures.
- This is the basis for more precise situation-specific relevance assessments.
- Attention data provides valuable knowledge for improved retrieval services
- as heuristics to partition documents
- as implicit and context-sensitive query extension and relevance feedback
- as additional information with retrieved documents (re-finding and re-presenting)
Mymory will deliver a retrieval model that integrates vector space models with fuzzy attention data.