Xapian devel - Nov 2014 - Suggested Sub-idea for Weighing Schemes[GSoC-2015]

The get_percentage() function brings us the relevance of document based on
terms but it is strictly based on a mathematical model rather than a
logical model which in all leads to false generation of priority of results
in an order.

What i am proposing is improvement of relevance of searches in documents
based on a partial mathematical and partial logical model,

For Example : Searching of Unique terms in a document or Matching number of
times the same query being searched being preferred more than the other
terms. Other could the the order of same query appearing in a document
whether appearing at beginning or in b/w or in end.

Mentors Give a feedback on this??

Regards,
Abhishek
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.xapian.org/pipermail/xapian-devel/attachments/20141125/237989e2/attachment-0002.html>

On Tue, Nov 25, 2014 at 04:05:41PM +0530, Abhishek Singh Kushwah
wrote:
> The get_percentage() function brings us the relevance of document based on
> terms but it is strictly based on a mathematical model rather than a
> logical model which in all leads to false generation of priority of results
> in an order.
You want to look at get_weight(), get_percentage() just returns the
weight scaled and rounded to give an integer score out of 100.  It's
really rather an archaic feature - showing percentage scores or star
ratings was popular for a while, but hasn't been for perhaps a decade.
> What i am proposing is improvement of relevance of searches in documents
> based on a partial mathematical and partial logical model,
What do you mean by a "logical model" vs a "mathematical
model"?
> For Example : Searching of Unique terms in a document or Matching number of
> times the same query being searched being preferred more than the other
> terms. Other could the the order of same query appearing in a document
> whether appearing at beginning or in b/w or in end.
Sorry, I don't follow what you're suggesting here.

Cheers,
    Olly