It would be super if you could agregate similarity object scores (preferably even weighted). For example

- Use the title score, weighted 0.8
- Add terms score, weighted 0.5
- Add Voting score, weighted 0.2

I think, in principle, the user is not really interested how you got the similarity, only if its relevant. Using agregated scores, the results become much more fuzzy, and with some proper trial-and-error can be much more interesting.

Comments

deekayen’s picture

Project: Similar Entries » Similarity Objects
Version: 6.x-1.0 » 6.x-1.x-dev
Scott Reynolds’s picture

Aggregate results are still push off for a while. Its actually pretty difficult to do using the current system. And im really not to keen on the idea.