Bug #5366
openIntegrate semantic query with keyword and spatial query
0%
Description
Combining semantic, keyword and spatial criteria will allow a smoother transition from non-semantic metadata to semantically annotated data. It will also allow us to refine query results further by limiting to specific organizations, locations, people, etc.
Metacat should support combining "squery", "semquery" and "spatialquery" actions in an extensible way that lets future query mechanisms contribute conditions to the query so that additional (or fewer) matches are returned.
This may require additional or wrapper syntax for specifying a compound, multi-faceted query.
Updated by ben leinfelder over 13 years ago
I've changed the DBQuery class in Metacat to allow combinations of a docid list (semantic and spatial queries generate these) with normal pathquery/keyword search criteria.
You can choose the normal mode (as it was before I made this change), or UNION mode or INTERSECT mode. The last two either expand or refine the search results.
The web UI still needs some work to incorporate these new features in an intuitive way.
Updated by ben leinfelder over 13 years ago
There a couple different behaviors possible for compound query techniques, and I'm not sure which we should choose or whether it should be user choice.
Expanding paradigm:
1. semantic concepts WITH keywords
-all docs matching the semantic concepts plus all docs that match the keywords
2. ONLY semantic concepts
-all docs matching the semantic concepts plus any other non-annotated docs (keyword contains "%")
3. ONLY keyword
-all docs matching the keywords (might still include annotated docs)
Refining paradigm:
1. semantic concepts WITH keyword
-docs matching the semantic concepts if and only if they match the keywords as well
2. ONLY semantic concepts
-only docs matching the semantic concepts (non-annotated documents are not returned)
3. ONLY keywords
-all docs matching the keywords (might still include annotated docs)
Updated by ben leinfelder over 13 years ago
from Shawn:
I vote for expanding, but with ranking.
This is like a google keyword search, where a match simply contains one of the search words, but docs with all search words are ranked higher (obviously the google algo is more complicated than this...but this is the general idea)
Updated by ben leinfelder about 11 years ago
- Target version changed from Unspecified to sms-index-1.0.0
Updated by ben leinfelder about 11 years ago
- Target version changed from sms-index-1.0.0 to metacat-plugin-0.9.0