Opened 13 months ago
Last modified 11 months ago
#7677 new enhancement
keep order for matches
| Reported by: | skyper | Owned by: | joshdoe |
|---|---|---|---|
| Priority: | normal | Component: | Plugin conflation |
| Version: | latest | Keywords: | match order |
| Cc: |
Description
I have a case where conflation should find better results. Seems to be a simple case as the subjects and the references are in inline order and this order should remain.
Thanks
Repository Root: http://josm.openstreetmap.de/svn Build-Date: 2012-05-05 01:31:01 Last Changed Author: simon04 Revision: 5211 Repository UUID: 0c6e7542-c601-0410-84e7-c038aed88b3b URL: http://josm.openstreetmap.de/svn/trunk Last Changed Date: 2012-05-04 19:15:59 +0200 (Fri, 04 May 2012) Last Changed Rev: 5211 Identification: JOSM/1.5 (5211 en) Memory Usage: 441 MB / 1820 MB (108 MB allocated, but free) Java version: 1.6.0_24, Sun Microsystems Inc., OpenJDK 64-Bit Server VM Operating system: Linux Dataset consistency test: No problems found Plugin: PicLayer (27883) Plugin: conflation (0.1.3) Plugin: jts (28210) Plugin: utilsplugin2 (28350)
Attachments (1)
Change History (6)
Changed 13 months ago by skyper
comment:1 Changed 13 months ago by skyper
- Type changed from defect to enhancement
comment:2 Changed 13 months ago by joshdoe
This seems to be a problem with OneToOneFCMatchFinder. I just tried DisambiguatingFCMatchFinder, which gives better results, but has one wacky match, which throws off the results. I think I'll have to create a new FCMatchFinder which uses the assignment problem solution (e.g. Hungarian algorithm). I have an implementation I can reuse in utilsplugin2, but sometimes it gets stuck in an infinite loop, so until that problem is resolved try v0.1.4 with the somewhat better finder (though essentially useless for your particular case).
comment:3 follow-up: ↓ 4 Changed 13 months ago by joshdoe
- Owner changed from joshdoe to skyper
- Status changed from new to needinfo
Something else that I could do is allow for marking some matches as invalid (rather than just removing them), and then re-running the matching process. If you have any suggestions as to how the UI should look for this let me know.
comment:4 in reply to: ↑ 3 Changed 13 months ago by skyper
Replying to joshdoe:
Something else that I could do is allow for marking some matches as invalid (rather than just removing them), and then re-running the matching process. If you have any suggestions as to how the UI should look for this let me know.
In generell, I am not that convinced with the two windows (toggle dialogue and seperate configure window). I think one should be sufficient but maybe I need to use it more often first.
It is always nicer if the machine can do the work but an option to manual interact is useful in tricky situtations.
comment:5 Changed 11 months ago by skyper
- Owner changed from skyper to joshdoe
- Status changed from needinfo to new



example file