Recently, looking at my author profile on MathSciNet (Am I a narcissist?), I saw that my citations counter is lower than what I expected. After a while, I realized what the problem is:

Many of my papers are cited (both by myself alone, by me plus some coauthors, or by other authors) as

"Authors names, Article title, Journal, (to appear)"

because they were accepted but not yet published at the time. Unfortunately MathSciNet does not count those citations.

Do you know how it is possible to fix this problem? Have any of you had the same problem?

Is there any official email address/website for suggesting corrections to MR citations? (quoting Sam Hopkins)

I guess it is a matter of contact in some way the MathSciNet's reviewers. I tried MathSciNet "Support Mail", but I guess it is only for browser/connection related problems, in fact I got no answers.

Thank you very much for any suggestion.


  • I found that when the "to appear" papers are cited as

"Authors names, Article title, Journal (year) (in press)"

then they are correctly handled by MathSciNet citations database. Thus I suggest to do so to everyone who has to cite a "to appear" paper.

  • This question is "on hold" since somebody thinks it is off-topic. I can agree that it is quite different from the typical MO questions. However, it seems to me that this kind of problem can interest many researchers, so I will wait at least until some satisfactory answer before close it.
  • $\begingroup$ I think this happens often and is hard to fix. Some papers do not appear, some papers appear in different journals than submitted to. Some papers differ greatly from arxiv versions. Google scholar seems to be able to collect a lot of these citations, but this probably does not help with job interviews... $\endgroup$
    – Thomas Rot
    Apr 6, 2016 at 9:19
  • $\begingroup$ @ThomasRot At least in my case, all the journal names are correct, because the papers were accepted (I never cite an article that has not been peer reviewed), what is missing is just the issue/year/pages information. $\endgroup$
    – Maxxt
    Apr 6, 2016 at 9:23
  • 3
    $\begingroup$ Google scholar is also not quite reliable unless the author himself update his reference list. Often Google Scholar is not able recognize that the same paper can be found from different sources (arXiv, journal website, different pre-print servers, the author's homepage, etc), resulting in multiple count of the same paper. On the other hand, if updated correctly, I find google scholar quite accurate. $\endgroup$
    – Raziel
    Apr 6, 2016 at 9:23
  • 6
    $\begingroup$ I guess it is a matter of contact in some way the MathSciNet's reviewers. I don't think that is the right way to proceed; bibliographic entries in Math Reviews are created by the editorial staff, not the reviewer. Moreover they are usually followed by some disclaimer like "This list reflects references listed in the original paper as accurately as possible with no attempt to correct error." That makes me sceptical that the editorial staff at MR would be willing to take on the task of updating these references. (I have sufferred from this too, and it is very frustrating!) $\endgroup$ Apr 6, 2016 at 10:09
  • 4
    $\begingroup$ A good practice in the information age would be that everyone who starts writing a paper generate a UUID for it like ac3e67f0-fbed-11e5-bc84-00221913161d and thereafter include it in all versions (preprint or published) and in all revisions, and similarly quote the UUID in all citations (the difference with a DOI is that it can be generated by the author, before publication). $\endgroup$
    – Gro-Tsen
    Apr 6, 2016 at 11:52

4 Answers 4


For citations from a reference list at the end of an article, we try to match items in the list to items already in the Math Reviews database. If a paper is "to appear" at the time it is cited, there is probably nothing in the database to match it to. We are currently developing a new matching program, one that should be better at matching papers that were cited as “to appear” but are now in the database. Once that is finished (and tested), it will replace the current method. We will also be running it regularly, so that we can find more matches of this type.

More information about citations in MathSciNet is available online at http://www.ams.org/mathscinet/help/citation_database_help_full.html#matched

If you have questions about the content of MathSciNet, including missed citations, you should write to [email protected]. Questions about the functioning of the site, such as problems with connectivity or math not rendering properly, you should write to [email protected]. If you are not sure which address to use, just pick one. The people who read the messages are good at triage. Or you can write to me: [email protected].

There is another type of citation in MathSciNet: from the body of a review of another paper. These can be particularly interesting, because a third party has decided that the two papers are related.

  • 1
    $\begingroup$ Thanks for posting this, and for your willingness to help! $\endgroup$ Apr 17, 2016 at 12:55
  • $\begingroup$ Any updates on whether this has been deployed? $\endgroup$ May 29, 2018 at 19:53
  • $\begingroup$ Yes, the new matching had been deployed. Nothing is perfect. So if we miss something, write to us at the email addresses given in the answer. $\endgroup$ May 29, 2018 at 19:56

Contact Edward Dunne: [email protected]

  • $\begingroup$ Ed Dunne is the User 49409. $\endgroup$ Apr 13, 2016 at 0:37
  • $\begingroup$ As in @EdwardDunne (just seeing if it works) $\endgroup$
    – David Roberts
    Apr 13, 2016 at 1:54
  • 4
    $\begingroup$ @DavidRoberts the '@' symbol is not a magic summoning command. If the user has not participated in a discussion already he/she would not be notified. $\endgroup$ Apr 13, 2016 at 19:50
  • $\begingroup$ Sure, but having a name might also be more useful/memorable to people than a user number, regardless of whether he was 'summoned' or not by @-mentioning him. $\endgroup$
    – David Roberts
    Apr 13, 2016 at 19:53
  • 4
    $\begingroup$ Here, let me add a comment, so you can @ me. $\endgroup$ Apr 13, 2016 at 20:48

My understanding is that "citations of preprints aren't counted" is the official MathSciNet policy. In fact, see
where it says "The Citation Database only includes citations from the reference list journals that are matched to items in the MR Database. Unmatched reference list entries are not included in any author or journal citation count."
I think that "matched" means "matched at time of processing the referring paper" here, although I'm not positive.

In general, after a couple of years, the MathSciNet citation counts are about as good a general measure of quality of/interest in a paper as any citation counts. But there is certainly a penalty for papers that spend a long time as preprints.

My opinion is that best practice is to report both. If I'm only going to report one, then I actually report the Google Scholar one. (But a few of my papers got a lot of citations while preprints.)

  • $\begingroup$ I can understand that preprints are not reported in citation. But in my question I'm asking about "to appear" papers, that is, articles accepted by a journal but without issue/year/pages yet assigned. $\endgroup$
    – Maxxt
    Apr 6, 2016 at 17:03
  • 1
    $\begingroup$ The "to appear" articles usually don't have a MathSciNet entry associated with them yet. I read the linked webpage as saying that having e.g. an MR number assigned as being the deciding factor as to whether or not the citation is assigned correctly. $\endgroup$ Apr 6, 2016 at 17:41

One point that was not mentioned in the comments above is that the database is periodically updated, so even if the article is currently not hyperlinked with a citation it may become so in the future. The mathscinet staff are generally helpful with links that are overlooked. Since nowadays this sort of "metric" is sometimes unfortunately a factor in promotion decisions, this is not an idle issue.

  • $\begingroup$ "The mathscinet staff are generally helpful with links that are overlooked." -> The problem is that I have to idea about how to contact MathSciNet staff to fix the reference. Do you know how to do so? $\endgroup$
    – Maxxt
    Apr 6, 2016 at 17:01
  • 1
    $\begingroup$ The email is [email protected], but you probably have already tried this, since this is the one that you get by clicking on "Support Mail" on MathSciNet. My experience with emailing this address in the past is that they will reply only under certain circumstances, when they feel that the issue that you are asking about is worth addressing. Otherwise, there will be no reply, and no action taken. I am almost sure that asking them to fix a very specific automatically-generated wrong or missing citation should be one of the circumstances when you do get a reply. $\endgroup$
    – YangMills
    Apr 6, 2016 at 17:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.