[Xapian-tickets] [Xapian] #804: Improve clustering API
Xapian
nobody at xapian.org
Thu May 30 03:53:49 BST 2024
#804: Improve clustering API
--------------------------+-------------------------------
Reporter: James Aylett | Owner: Olly Betts
Type: enhancement | Status: assigned
Priority: highest | Milestone: 1.5.0
Component: Library API | Version: git master
Severity: normal | Resolution:
Keywords: | Blocked By:
Blocking: | Operating System: All
--------------------------+-------------------------------
Comment (by Olly Betts):
Checking the paper, it seems we should preserve the relative order of
'''promoted''' documents (which I'm fairly sure isn't the case currently
in our implementation).
It seems the expectation is that only the promoted documents ("promoted"
is my terminology, not used in the paper) are returned as the final
results, but in Xapian I think we want to keep the non-promoted documents
in the results since then it's possible to more simply implementing paging
through all the results (you could diversify the top 100, then page on
from there as you would in the undiversified case). So we effectively
want to do a stable partition of the MSet such that all promoted documents
are before all non-promoted ones. We may even be able to implement that
in-place!
--
Ticket URL: <https://trac.xapian.org/ticket/804#comment:5>
Xapian <https://xapian.org/>
Xapian
More information about the Xapian-tickets
mailing list