button to remove self-citations

[Christophe]

Hello,

Sorry if the suggestion has already been done but I don not see it :

Would be useful to have a button option to remove self-citation or
auto-citations, at least direct one (the quoter is in the authors
list). Self citations completely invalidate computed index numbers.
All papers have self citations but for some papers it can go to 60% or
70%.

Congratulations and thanks for the tool.
Cheers
Christophe

[Fabrizio Sebastiani]

Yeah, that's a great suggestion! I guess that the definition of self-citation should be "a non-null intersection between the author list of the citing paper and the author list of the cited paper".

Fabrizio

> --
> You received this message because you are subscribed to the Google Groups "Scholarometer" group.
> To post to this group, send email to schola...@googlegroups.com.
> To unsubscribe from this group, send email to scholaromete...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/scholarometer?hl=en.
>

[Christophe]

On Jan 25, 3:42 pm, Fabrizio Sebastiani <fabse...@gmail.com> wrote:
> Yeah, that's a great suggestion! I guess that the definition of self-citation should be "a non-null intersection between the author list of the citing paper and the author list of the cited paper".

Yes that's what I meant, roughly expressed. Thanks.

Cheers
Christophe

[Fil Menczer]

Thank you for the great suggestion! Of course, it is not trivial to
implement, as currently we get the citation counts with just one (or a
few) google scholar queries, whereas to check for self-citations would
require (at least) one query for each paper. This would slow things
down *a lot*. I suspect this is also the reason why other tools that
rely on google scholar do not filter self-citations either, as far as
I know...

Perhaps this could be an option for people who are willing to wait a
lot of time. We added it to our (growing) to-do list, and will get to
it as resources permit.

Thanks,
-Fil & Scholarometer Team
http://cnets.indiana.edu/people/filippo-menczer

[Bernard Silverman]

Of course, it is understood when looking at indices based on citations
that self-citations are included. Mostly, this is unlikely to make much
difference because self-citations tend to be of two, entirely
legitimate, kinds:

1. citation of a rather obscure paper which gives further details of
work in the main paper, which might be published in a journal like
Nature or Science which only publishes brief papers; so this would add a
small number of citations to papers not much cited

2. citation of papers already well cited, which anchor the citing paper
in the general literature; so this would slightly increase the number of
citations to papers already highly cited

Apart from the fact that these self-citations could be regarded as
entirely legitimate...ie would and should have been made regardless of
the authorship of the various papers...they are unlikely to affect
h-indices materially, because to do that you have to generate citations
to papers which are in the "middle" category. What one would wish
to exclude are "gratuitous" or even deliberately manipulative
self-citations.

All I would say is that this matter is much more subtle than one would
think. Of course, if Scholarometer could provide detailed analyses of
self-citations, that might open up the possibility of further
investigating the phenomenon of self-citation, but I can see why Fil and
team are somewhat hesitant to devote large resources at this time.

Bernard

On 26/01/2011 00:02, Fil Menczer wrote:
> Thank you for the great suggestion! Of course, it is not trivial to
> implement, as currently we get the citation counts with just one (or a
> few) google scholar queries, whereas to check for self-citations would
> require (at least) one query for each paper. This would slow things
> down *a lot*. I suspect this is also the reason why other tools that
> rely on google scholar do not filter self-citations either, as far as
> I know...
>
> Perhaps this could be an option for people who are willing to wait a
> lot of time. We added it to our (growing) to-do list, and will get to
> it as resources permit.
>
> Thanks,

> -Fil& Scholarometer Team

[Christophe]

On Jan 26, 1:02 am, Fil Menczer <f...@indiana.edu> wrote:
> Thank you for the great suggestion! Of course, it is not trivial to
> implement, as currently we get the citation counts with just one (or a
> few) google scholar queries, whereas to check for self-citations would
> require (at least) one query for each paper. This would slow things
> down *a lot*. I suspect this is also the reason why other tools that
> rely on google scholar do not filter self-citations either, as far as
> I know...
>
> Perhaps this could be an option for people who are willing to wait a
> lot of time. We added it to our (growing) to-do list, and will get to
> it as resources permit.
>
> Thanks,

Dear Fil,
In my very naïve view of the implementation. I thought that you had
the result of queries stored in
some kind of collection of your prefered language. If true, it could
then be somehow not too hard to locally apply the "remove-self"
algorithm before displaying the collection without doing other
queries. I thus understand it is not so simple ...
Indeed I have seen no other tool that remove self-citations.
Thanks for your efforts.
Cheers
Christophe Dony

[Christophe]

On Jan 26, 7:14 am, Bernard Silverman

<bernard.silver...@stats.ox.ac.uk> wrote:
> Of course, it is understood when looking at indices based on citations
> that self-citations are included.  Mostly, this is unlikely to make much
> difference because self-citations tend to be of two, entirely

> legitimate, kinds: ....

Of course I entirely agree that self-citations are perfectly
legitimate, I dont say people should not make them in their papers.

>
> Apart from the fact that these self-citations could be regarded as
> entirely legitimate...ie would and should have been made regardless of
> the authorship of the various papers...they are unlikely to affect
> h-indices materially,

I do not really agree. Let me exclude from the discussions those top
researches with hundred of citations, self citations make no
difference for them; but I can now see that indexes are used
everywhere and also for phd students, post-docs, tenure etc.
Even if, of courses, indexes are just one indicator among many others,
to my opinion having a paper with let's say 20 "true" (non self)
citations by let's say 10 different quoters, is a proof that the work
has had an influence at its level.
It is somehow different if 15 or more of these 20 are self-citations
(really not difficult to find examples of this), especially with that
todays tendancy to publish so many papers.

> All I would say is that this matter is much more subtle than one would
> think.  Of course, if Scholarometer could provide detailed analyses of
> self-citations, that might open up the possibility of further
> investigating the phenomenon of self-citation, but I can see why Fil and
> team are somewhat hesitant to devote large resources at this time.

No problem, I understand it takes a lot of time, it was just a
suggestion.
Thanks for the discussion and again congratulation to Fil and team.
Cheers

[Fil]

Thanks everybody for the helpful suggestions, comments and discussion,
much appreciated.

We would not want to make the filtering of self-citations the default
behavior (for the reasons clearly articulated by Bernard) but if we
have time and can implement it efficiently (something I am not sure
of) it would certainly be a nice option for the advanced interface.

-Fil & Team

[Axel Brandenburg]

Dear Developers,
Excluding self-citations is required for applications to the European Research Council
and the deadline is 22 November. One can try it by hand, counting off the self-citations,
but that is not very reliable. So, a non-default option, even if it slows down the search
a lot, would help!!
Thanks, Axel

[Fil Menczer]

Dear Axel, this feature is in our to-do list.

Thank you,
-Fil

> --
> You received this message because you are subscribed to the Google Groups
> "Scholarometer" group.

> To view this discussion on the web visit
> https://groups.google.com/d/msg/scholarometer/-/o7s1CS2_4FQJ.

[Axel Brandenburg]

Hi Fil,
Sounds great. Do you think you'd be able to use me as
a test case. I just need the number of citations (without
self-citations) for 10-11 papers of the last 10 years:

http://scholar.google.se/scholar?oi=bibs&hl=en&cites=4545663113421842523
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=9776582557700616280
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=4840075774206831835
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=7714423359191427490
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=2598949910401878941
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=9573997055053225610
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=11450325191180685575
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=4414803312053590121
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=16605869972830735808
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=14258396345344756608
http://scholar.google.se/scholar?oi=bibs&hl=en&cites=15634445393995433375

It would help if I could get *all* (not just 20) papers on one screen,
and then I could just count off the self-citations.
Axel

> >> > In my very na�ve view of the implementation. I thought that you had

[Fil Menczer]

As explained in a previous email, we currently do not have support for
this project, so our progress is slow. There are other more pressing
features to be address first, based on ongoing work and students'
research plans. The time scale will be (possibly many) months, just so
you do not hold your breath.

-Fil

PS - Of course, if a particular feature is sufficiently important to
attract funding, priorities can be adjusted. Eg, a grant of about
US$80k would allow us to hire a full-time programmer and thus revisit
our priorities ;-)

>> >> > In my very naïve view of the implementation. I thought that you had

[Johan Pouwelse]

Just a +1 for this feature.

My research focus is reputation systems and measurements of Internet-deployed systems.

A small minority of academics in the bottom to middle category have an h-factor

which relies mostly on self-citations, according to my scholar.google sampling.

Obviously an extremely sensitive matter, considering tenure track decisions and 

general scientific standing...  -j

[Csaba Szepesvari]

This would be very useful not only for people submitting grant applications, but also for reviewers.
Cs.

[Fil Menczer]

Dear Christophe,

Thank you for the note. Indeed this feature has been on our to-do list
for a long time. Unfortunately, we do not have external support for
the project and therefore we are able to add features at a very slow
pace. Therefore we are unable to make any promises as to when we will
be able to get this done. Currently we are working on other
improvements that are even higher priority.

Thank you for your support and interest in the project!

-Fil & Scholarometer Team

> --
> You received this message because you are subscribed to the Google Groups
> "Scholarometer" group.

> To unsubscribe from this group and stop receiving emails from it, send an
> email to scholaromete...@googlegroups.com.

> To post to this group, send email to schola...@googlegroups.com.

> Visit this group at http://groups.google.com/group/scholarometer.
> For more options, visit https://groups.google.com/d/optout.

[Jing Jiang]

How time flies! Just curious if this function has some progress over the last 2 years?

在 2014年3月9日星期日 UTC-5上午11:52:54，Fil写道：

[Fil Menczer]

Sadly I must report that our resources have shrunk rather than
increase, and so we have no updates on this.

-Fil

> Visit this group at https://groups.google.com/group/scholarometer.