#420: Citesummary quirks
-----------------------+----------------------------------------------------
Reporter: tbrooks | Owner:
Type: defect | Status: new
Priority: major | Milestone:
Component: WebSearch | Version:
Keywords: INSPIRE |
-----------------------+----------------------------------------------------
There are 3 quirks/improvements to be done with citesummary format:
1) The links on the numbers in the citesummary tables now function
correctly if the users uses advanced search, as per #340. However, the
links are generated by taking the users search and appending one or two
terms in invenio syntax (i.e. user searches
{{{
Author:brooks
}}}
and citesummary generates
{{{
author:brooks and Collection:citeable and cited:50->100
}}}
This is fine. But Invnenio syntax mixed with SPIRES syntax can be odd,
and many (most?) people will be doing:
{{{
find a brooks, t
}}}
so the resulting links are
{{{
find a brooks, t and collection:citeable and cited:50->100
}}}
I have not seen any evidence of this failing, but mixing the syntaxes can
lead to odd results, especially with authors. This should be confirmed
to work correctly and/or fixed to ensure it works.
2) We should add a link to "remove Review of Particle Properties from
analysis" (we might check if one is in the results first) We usually do
this by {{{
-title:rpp
}}}
since anything with title:rpp is assumed to be an RPP and thus off the
carts as far as citation analysis goes. People like to be able to remove
these easily and redo the analysis. We might check the result for
possible RPPs and then display "This set includes at least one Review of
Particle Properties, which is usually very highly cited, click here to
remove from analysis"
3) When one follows the links off of the numbers, the results are sorted
in "times cited order" I think this is incorrect, and confusing to the
user - particularly if one then wants to do another search using the
regular search box. It is too hard to see that the search order has
been changed, and the new search is in a confusing order. It is
arguable whether the search from the citesummary link should be in times-
cited order. However, the fact that the users then needs to change the
rank order (which is hard to find...) back to "or rank by" (which is not
intuitive...) to get the standard search behavior is a bad problem,
especially for the following use case:
Search a set of papers (say by a particular person)
Look at the top to see what they've done recently
Select citesummary to see their overall picture
Look at the few highly cited ones to see what they've done that was big
(here sort barely matters)
Now, you have a good picture of the person, so you move to the next
(reviewer/candidate/author/friend/whatever):
Search the set of papers
Look at the top to see what they have done recently....
oops, the top is now the highly cited papers by this person, how did that
happen...
So I suggest that the links off the numbers should report in default sort
order - by the principle of least surprise. possibly one could solve this
by other means as well.
see also #326 #134 for other changes to this popular format.
--
Ticket URL: <http://invenio-software.org/ticket/420>
Invenio <http://invenio-software.org>
