I think you're going to have trouble doing this with seperate cores.
With seperate cores, you'll need to issue two querries to solr, one for
each core. And then to intermingle results from the differnet cores like
that, it's going to require difficult (esp to do at all efficiently)
client side code. Different cores really are entirely seperate.
If you put everything in the same core, but with different solr fields
used, it will be easier to intermingle. Conceptually, think of all the
documents you were thinking of as being indexed in 'core1' as being just
in the single core, but having their text indexed in a field called,
say, "text_core1". Then the 'core2' documents are in that same core
actually, but indexed under "text_core2".
Now if you just wanted results from core1 followed by results from
core2, you could use dismax and boost the core1 field a lot, something like:
text_core1^1000 text_core2
Then we add in your requirement for "exact matches" first. You need to
be more precise about what you mean by "exact" matches and "partial"
matches. By "exact" do you mean phrase searching? Do you mean it must
match the entire field exactly start to finish? Do you mean un-stemmed,
where "partial" is stemmed?
Once you figure this out, one possible way to approach it is to set up a
solr field with analyzers such that it will only match "exact" matches.
For instance, if you really mean exact string match without any
tokenization, you could use the KeywordTokenizer. If you are able to set
up a solr field like this, then again using dismax, the solution is
straightforward, something like:
qf=text_core1_exact^3000 text_core2_exact^2000 text_core1_partial^1000
text_core2_partial
Hope this helps you think about how to approach your problem.
Jonathan
Balaji.A wrote:
Hi All,
I have a specific requirement as stated below. Kindly suggest if this can
be acheived or not and the steps to acheive it.
I have 2 cores storing different kind of data.
My search query should return results in the below given order
1) Exact match resutls from core1
2) Exact match results from core2
3) Partial match results from core1
4) Partial match results from core2
Note: I don't want exact match results to be duplicated in Partial match
results.
Please suggest!
Thanks.
