Searching Data Wells

Members:
  • Mats Almkvist
  • Linus Ericson
  • Markus Sköld
  • Vaclav Rosecky
  • Søren Mollerup
  • Tine Musak

Integrated search: Does it even make sense with very large amounts of heterogeneous data objects?
Segmented search: where to slice it?
Enhanced search: Enhance with what?

1. Required metadata
    Is it necessary to require a minimum set of metadata to be able
    to search the content of a stored object?

2. Analyzing Object types
    Extra metadata can be extracted from objects if the format is
    known, but this requires some <analyzing of the specific object
    type, which have to be done by hand by people who know the
    format. is this worth it?

3. Segmented search
    - Boosting at indexing time  
      Are some objects always more important than others? ie. should
      they always be on top of a search result?
   
    - Boosting at search time
      If the Results can be boosted at search time it makes it
      possible to boost certain objects, based on who performs the
      search. Could this be used to make searches even more
      relevant?

4. Enhanced Search
    - There are many posibilities to enhance the search results.
      Relations between objects in the repository could provide extra information 
      to the search. For example if an object is part of a collection.
    - Facets could provide extra information about the objects in
      the repository in relation to a specific search query.
    - Other features like auto-suggest also enhances the user experience.