TF/IDF should be pretty simple to implement IMO.
And no, Solr does not give preference to prior documents.
However, Solr allows you to "boost" specific terms, thus creating the impression of preference.
On Fri, 4 Mar 2022, 22:15 Bruce Momjian, <bruce@xxxxxxxxxx> wrote:
On Fri, Mar 4, 2022 at 11:43:57AM -0500, Tom Lane wrote:
> "Bayer, Samuel" <sam@xxxxxxxxx> writes:
> > One concrete question, I suppose, is: the classic TF/IDF search strategy relies on inverse document frequency, which looks across the corpus. I can't tell whether that corpus-wide frequency information is taken into account in either ranking function.
>
> The documentation is pretty clear that they don't, they just consider each
> document in isolation. Building a structure that would allow more-global
> info to be taken into account is an interesting project that nobody's
> tackled.
Oh, you mean like, "pizza" is used only in this document vs it being
used in every document. Should we add some documentation about this or
is it already clear?
--
Bruce Momjian <bruce@xxxxxxxxxx> https://momjian.us
EDB https://enterprisedb.com
If only the physical world exists, free will is an illusion.
