8
$\begingroup$

Are there policies to respect in order for MathOverflow Q&As to be tracked back in relevant zbMATH Open reviews?

I am asking this since I noticed that all questions and answers to which I contributed, fully citing sources with their MR and/or Zbl reviews as it is my custom to do so, have all been tracked back inside the relevant zbMATH Open reviews except for this latest Q&A: I noticed also that other users' answers were tracked back, so I wonder if it was so because I didn't respect a policy (I'm unaware of).

$\endgroup$
11
  • 5
    $\begingroup$ I have to admit that it's not clear to me what is actually the question. (To me it sounds like a question about some kind of trackbacks - but I was unaware of such feature in ZentralBlatt.) Still, I will mention that there is a tag (citations) here on meta - perhaps looking in that tag you might find some related questions. (And it might be a suitable tag for this question, too.) $\endgroup$ May 7 at 14:56
  • $\begingroup$ Thanks @MartinSleziak, I added the tag you suggested $\endgroup$ May 7 at 15:01
  • 1
    $\begingroup$ So just to clarify (since for me it was unclear before I saw an example): When I look at Zbl 0384.26002 on their website, I see there a link: "MathOverflow Questions: Generalisation of Cauchy’s mean value theorem". So basically you're asking under which conditions the link back to MO appears on the Zentralblatt's website, right? $\endgroup$ May 7 at 15:33
  • $\begingroup$ @MartinSleziak, yes. Nearly few months ago, I noticed the linking back from zbMATH Open, and I wonder why it as not worked for my answer to that question. $\endgroup$ May 7 at 15:36
  • 2
    $\begingroup$ This may simply be an issue with the specific URL you used to cite zbMATH, i.e., zbmath.org/?format=complete&q=an:0089.19103 instead of zbmath.org/?q=an:0384.26002. It may be the case that only the latter form is detected. $\endgroup$ May 7 at 20:02
  • $\begingroup$ Thanks @DmitriPavlov: I'll try to change the seemingly "ill" defined links. $\endgroup$ May 7 at 20:05
  • 1
    $\begingroup$ @DmitriPavlov and Daniele Tampieri: I found 8 posts containing links in the other format, the complete list is here, one can use this search. I did not go through all of them - but in the ones I checked, there's isn't a trackback either. $\endgroup$ May 8 at 14:02
  • $\begingroup$ @MartinSleziak I don't yet have the full proof that Dmitri has solved the problem, but I'll try to fix all of them following his suggestion, and let's see. $\endgroup$ May 8 at 14:23
  • 1
    $\begingroup$ On a different but related note, it seems that trackbacks are also not recorded wherever the old URL zentralblatt-math.org was used. $\endgroup$ May 9 at 19:30
  • $\begingroup$ @TheAmplitwist: thanks for pointing it out. I just noticed that the trackback now works for the first link I modified according what Dmitri Pavlov's findings: from now on I'll change also the old URL links. $\endgroup$ May 10 at 19:20
  • $\begingroup$ @DmitriPavlov, if you'll change your comment to an answer, I'll upvote and accept it. $\endgroup$ May 10 at 19:23

2 Answers 2

9
$\begingroup$

This is an issue with the specific URL used to cite zbMATH, i.e., https://zbmath.org/?format=complete&q=an:0089.19103 instead of https://zbmath.org/?q=an:0384.26002. Only the latter form is detected. Also, the domain name has to be zbmath.org, not the synonymous zentralblatt-math.org.

$\endgroup$
23
$\begingroup$

I am in charge of the MathOverflow backlinks at zbMATH Open and can explain how we fetch them:

We are using a small Python script, which uses the Stack Exchange API, to get all question, answers and comments citing some zbmath document on mathoverflow.net. We use urlextract and regular expression to extract links to zbmath.org.

The script runs once a day. The links that were found are inserted into our database and displayed at zbmath.org with a delay of usually one day. So in total there is a delay of two days from linking to a zbmath document in a MathOverflow post and the display of the backlink on zbmath.org. (It is possible to reduce this delay. We thought this is not needed.) Revisions of old posts are also considered.

Once in a month we run a script to remove invalid MathOverflow links or backlinks to MathOverflow Posts, in which the corresponding zbmath link was deleted. It would also be possible to run this deletion script every day. As we observed that it does not happen that often that backlinks have to be removed, we decided to run it only once in a month. For more details, see

Müller, Fabian; Schubotz, Moritz; Teschke, Olaf, References to research literature in QA forums – a case study of zbMATH links from MathOverflow, Eur. Math. Soc. Newsl. 114, 50-52 (2019). ZBL1428.68357.

As @DmitriPavlov and @TheAmplitwist noticed, links with ?format=complete or using the old zbmath domain zentralblatt-math.org are not recognized by our script. This is a bug that we will correct.

Regarding the original question: There is no policy on formatting links in order for MathOverflow posts to be tracked back in relevant zbMATH Open reviews. We recommend to use the “Insert Citation” button (only available at mathoverflow.net, not on meta.mathoverflow.net). Using this button one can search for a citation string and gets a list of up to three relevant zbmath documents. If the correct document is among the three results one can choose the correct one and the citation will be inserted correctly formatted into the MathOverflow post. However, also links that were pasted by hand should work as long as the zbmath identifier (zbl number or jfm number) occurs in the link. For example,

https://zbmath.org/?q=ti:References+to+research+literature+in+QA+forums

does not work, but

https://zbmath.org/?q=an:1428.68357

does work.

To make sure the link is recognized, I would advise to use the format zbmath.org/?q=an:<zbl id>. It is also possible to link to several zbmath documents by separating the individual zbmath identifiers using ‘|’. For example:

https://zbmath.org/?q=an:1428.68357|1422.01015

If some backlinks do not occur on zbmath.org after two or three days, you can write a mail to editor@zbmath.org and we will investigate the problem.

Furthermore, we are working on linking to more websites that are interesting for mathematicians. We started with the Digital Library of Mathematical Functions and will include The On-Line Encyclopedia of Integer Sequences (OEIS) in a next step. If you want other websites or services to be included, you can also write to editor@zbmath.org.

$\endgroup$
2
  • 6
    $\begingroup$ Thanks for reaching out to the MathOverflow community, it is certainly appreaciated! $\endgroup$ May 12 at 9:25
  • 2
    $\begingroup$ Thank you so much Isabel. I apologize for my later in thanking you earlier, but I think that @MartinSleziak did a good job in his comment. I'd also like to point out that I accepted Dmitri's answer not because your own is worser, but because he has been the first to realize where the problem was: I consider both your answers as a solution to the problem I posed. $\endgroup$ Jun 4 at 21:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .