Analyzing SPIRES Database for String and Non-String QG Cites

[marcus]

a simple way to check how things are going with string and non-string QG is to look at the list of last year's topcite papers. Doing this for the first time just establishes a baseline for comparison later. I don't expect anything surprising to come out of it----string research is still dominant. (the following would indicate by a factor of 2) I'll total the cites of top papers in several categories.

For some people this could be an introduction to using SPIRES database. You go to Spires and type in the following search terms.

FIND DK STRING MODEL AND DATE = 2006
Total top ten 949. Average cites per paper 94.9

FIND DK LOOP SPACE OR SPIN, FOAM AND DATE = 2006
Total top ten 437. Average cites per paper 43.7

It doesn't seem worthwhile looking at DATE = 2007, because papers with publication date 2007 have hardly had a chance to be cited yet. The other researchers have still to assimilate and utilize whatever is there. The search terms "loop space or spin, foam" succeed in picking out papers in LQG, Loop Quantum Cosmology, and Spinfoam. But one would not expect them to bring up papers by Loll (triangulation) or Reuter (asymptotic safety). I am still trying various search terms to get other non-string QG.

You might be curious to see what the LQG and Spinfoam papers are that are brought up. Here is a link for that, where the list is ordered by approximate cites
http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+DK+loop+space+or+spin%2C+foam+AND+DATE+%3D+2006&FORMAT=www&SEQUENCE=citecount%28d%29

 

[Coin]

What would be interesting to me, although I suspect the simple SPIRES search interface could not do this, would be to see whether there's any appreciable degree of cross-citation between those two groups of string-based and "geometric" (for lack of a better word-- i.e. loop, spinfoam, triangulation-based) QG papers. I.E., if you make a big citation graph of last year, and look at the citation "children" of the string papers and the geometric qg papers, do they form more or less exclusive sets? Or are there any appreciable number of papers that cite both a "geometric qg"-clique paper and a string-clique paper?

I'd also be a little curious whether the strength of the string search changes if you exclude papers about ads/cft (or does the use of "string model" do this already?)

 

[marcus]

do they form more or less exclusive sets? Or are there any appreciable number of papers that cite both a "geometric qg"-clique paper and a string-clique paper?

There is an outfit that makes these citation graphs showing clustering in ArXiv preprints. Arivero here at PF has sometimes given links to their analysis. It is based in Russia and I think they identify fields of research by means of cluster analysis and then collect stats on these (objectively determined) areas of research. All or mostly automatic. I can't think where to find the links at the moment. Maybe someone will notice your question and help out.

My subjective impression is that citation is clubby within the separate communities. Loop conferences, however, do tend to invite one or more plenary speakers from several other fields including string.

I'd also be a little curious whether the strength of the string search changes if you exclude papers about ads/cft (or does the use of "string model" do this already?)

Coin, you could help considerably if you would experiment with Spires search terms and see if you can improve on what I'm using. With my limited understanding of their keyword system, there is no substitute for actually trying and seeing what top ten papers come up. Just go here
http://www.slac.stanford.edu/spires/
and type in
FIND DK STRING MODEL AND DATE = 2006
and what you get is
http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+DK+STRING+MODEL+AND+DATE+%3D+2006&FORMAT=WWW&SEQUENCE=citecount%28d%29

(I selected the "approximate cite count" ordering of the list)

The top ten are roughly half AdS/CFT, judging by title, so I would GUESS that just searching by "string model" does an OK job of digging up AdS/CFT, at least highly cited ones. But you should check for yourself to be sure.

========================
EDIT TO REPLY TO NEXT
Coin, thanks for taking a look. Let me know if you find terms that dig up more highly-cited stringy papers!
I am not aware of any offer by Spires of a downloadable version. But their site has a lot of features including a kind of Q/A chat area, as I recall. They also ask for suggestions and feedback. So if you ask a question there is probably somebody alive who will answer. The librarians who do the work of tagging the records with keywords are at DESY in Germany. (they also extract keywords automatically from title, but it is not so effective, I find)
The DESY librarians are not able to stay caught up, naturally. Sometimes they seem about a year behind. So that is another reason to just look at 2006 papers instead of trying to rank 2007 papers (the engine doesn't even know they re there yet in some cases). For a long time they didn't have much in the way of non-string QG-----you couldn't just say "loop space" and get a representative bunch of stuff. But in the past 2 years it seems to have improved. On the whole I think Spires is great, especially now.

 

[Coin]

I can try to give it a look later, thanks.

Incidentally, do you know, does SPIRES or any similar service offer any kind of downloadable version of their citation index, such as could be used for offline data analysis?

 

[marcus]

I can try to give it a look later, thanks.

Incidentally, do you know, does SPIRES or any similar service offer any kind of downloadable version of their citation index, such as could be used for offline data analysis?

Unfortunately I don't. I replied in part in the preceding post. But in case you or anyone else is interested in raw research output here are some from Spires by year to check for trends. one column is what the DESY librarians tagged "string model" and the other is what they tagged "loop space or spin, foam"

What they have so far for 2007 (in parens) doesn't mean much because it always takes them months to get thru the year's research publications.

        st  lqg/sf
1991   766   46
1994   774   65
1997   960   45
2000   949   64
2003   857   74
2006   721  100
2007   (89) (39)

 

[marcus]

More yearend stats.
A search of the Harvard database for 2007 papers using five keywords:
superstring, heterotic, M-theory, brane, AdS/CFT
http://adsabs.harvard.edu/cgi-bin/n...txt_wgt=YES&ttl_sco=YES&txt_sco=YES&version=1
came up with 852, this may increase over the next couple of weeks due to late entries.

There has been a decline in the number of research papers using those five keywords in their abstract summaries. Here are some sample years

2002: 1155
2006: 977
2007: 852 (may increase after close of year)

What I'm seeing is an average decline of 50/year for the years 2002-2006
and a more rapid decline (twice as fast, around 100/year) recently.
It is not clear what this means or even if it is significant

==================

Here's a graphic chart of ArXiv monthly submission rates thru end 2007 in broad categories like
Astro, Condensed, Hep,...
http://arxiv.org/Stats/hcamonthly.html

It looks to me as if Astro is still climbing.
Condensed may still be on an uptrend or may have plateaud
Hep has plateaud or may be slightly declining.*

*this is curious because Hep is getting a big influx from Asia now. So since the total Hep is roughly steady there must be a decline in productivity of the Yankee-Euro contingent-----or a temporary lapse while people are shifting from one specialty to another. String output has been declining, so maybe people are getting out of string and into other lines of research (Astroparticle, Cosmology...?) and the inconvenient hassle of making a move out of string has temporarily slowed them down.

===================
Much of the QG research I follow posts in the GR-QC part of ArXiv. this is a rather small part of the total, only about 4 percent, with monthly submissions only about 100 per month. However, in case anyone wants the chart for that piece, here it is
http://arxiv.org/Stats/gr-qc_monthly.png

It looks to me as if there is still slow steady increase in gr-qc (general rel and quantum cosmology). the slope is not much different from what it was 5, 10 or 15 years ago

 

[marcus]

In case we want to update the yearend stats, I will keep the links handy

FIND DK STRING MODEL AND DATE = 2006
721 papers, top ten got 949 cites--average per paper 94.9
http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+DK+STRING+MODEL+AND+DATE+%3D+2006&FORMAT=WWW&SEQUENCE=citecount%28d%29

FIND DK LOOP SPACE OR SPIN, FOAM AND DATE = 2006
100 papers, top ten got 437 cites--average per paper 43.7
http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+DK+loop+space+or+spin%2C+foam+AND+DATE+%3D+2006&FORMAT=www&SEQUENCE=citecount%28d%29 [/QUOTE]

The cites were counted at yearend 2007 (27 December), and would most likely have increased since.
Doing the same two searches for successive years to spot trends, one gets these yearly publication figures:

        st  lqg/sf
1991   766   46
1994   774   65
1997   960   45
2000   949   64
2003   857   74
2006   721  100

This compares total numbers of publications tagged "string model" with those tagged "loop space or spin, foam".

We can update the Harvard abstracts data base figures. Publications with any of the five keywords (heterotic, superstring, M-theory, brane, AdS/CFT). There has been a decline since 2002. As of today the totals for three sample years are
2002: 1171
2006: 978
2007: 915

More links can be found in previous posts on this thread.