PIM/Akonadi/SearchInfrastructure

All about the search infrastructure of Akonadi.

Concept

Concept of the Akonadi search infrastructure:

Query Language

Options:

• SPARQL: de-facto standard by now (used by Nepomuk and Tracker), but way too complex for automatic query transformations.
• XESAM (?)

Search Backends

Requirements:

• fast enough
• cross-platform
• persistent/live search with change notifications

Possible Backends

• Nepomuk with Sesame2 or Virtuoso
• Tracker (?)
• XESAM (?)

Delegation to Resources

• Requires query transformation
• Requires management for live searches
• Requires the ability to report search results (needs protocol extension) [done by now]

Virtual Collections

Virtual collections are used to report results, for details see PIM/Akonadi/VirtualCollections.

Nepomuk Feeder Agents

• The delete/add approach removes otherwise added information such as tags when an item is modified.
• Category to tag conversion uses non-Fast Nepomuk resources.