15

I've noticed, for citations, that some BibTeX files provided by google scholar are having regular errors (I believe these are errors).

For bib items, there is the "pages" field, in which we could input a range or a number of pages an article has. I've found at least two journals (in a short time span) whose articles have a wrong "pages" field. For instance

@article{baffou2009heat,
title={Heat generation in plasmonic nanostructures: Influence of                  morphology},
  author={Baffou, G and Quidant, R and Girard, C},
  journal={Applied Physics Letters},
  volume={94},
  number={15},
  pages={153109},
  year={2009},
  publisher={AIP}
}

or

@inproceedings{ammari2015super,
  title={Super-resolution in high-contrast media},
  author={Ammari, Habib and Zhang, Hai},
  booktitle={Proc. R. Soc. A},
  volume={471},
  number={2178},
  pages={20140946},
  year={2015},
  organization={The Royal Society}
}

Similarly, this last paper is not a proceeding, but an article from the journal "Proceedings of the Royal Society".

Probably there are some other mistakes around.

I know that these data are collected automatically, but, is there any way to report the errors to help google scholar to perfect their algorithms?

closed as off-topic by Coder, Solar Mike, user3209815, Buzz, Fomite Mar 14 '18 at 23:51

  • This question does not appear to be about academia within the scope defined in the help center.
If this question can be reworded to fit the rules in the help center, please edit the question.

  • Look it is done by a program. There can be bugs. Simple, correct and use it as per your requirement. Change inproceedings to article. – Coder Mar 14 '18 at 13:52
  • 5
    I'm voting to close this question as off-topic because this is not particularly about academia. It is about a bug in some third party website. – Coder Mar 14 '18 at 13:52
  • 12
    Yes, but it is a useful tool in academia, as we all should be citing properly. And yes, you could just cross your arms and do nothing about a systematic mistake or seek a way to make our life's easier; there is even a google scholar tag for the questions. In case it is off topic, it begs the question, where could I ask something like this? – pancho Mar 14 '18 at 13:55
  • 15
    @Coder Sure, there can be bugs, but bugs can be corrected, and I think complaints from users are one of the main factors in decisions about which bugs are worth correcting. – Andreas Blass Mar 14 '18 at 13:58
  • @FranciscoJoseRomeroHinrichs There can be error in publisher's metadata. For example, one of my paper show the year 1926, another shows 1845. This is strange. Google scholar once replied asking me to contact publisher. Publisher said everything is okay. I gave up. – Coder Mar 14 '18 at 19:03
12

The obvious place to report it would be the Google Scholar "contact us" form. I suspect there is little chance of the problems being fixed (since scholar has been around for over a decade, and I'm sure others must have reported these problems over the years) but you have little to lose by trying.

  • 2
    Thank you, I couldn't find this. You could be right tho, but I think it is worthy a try. – pancho Mar 14 '18 at 15:14
  • As a software developer (not with Google), yes, please report bugs like this. You may not see the bugs fixed, and they're probably not going to tell you, but they can't fix bugs if they don't know they're there, and by reporting the bug, you're helping to make sure that they do. – Nic Hartley Mar 14 '18 at 21:30
13

I am an engineer on the Google Scholar team.

Errors in citation data provided by Google Scholar can occur either because of an error at the source or a case that is not handled correctly by our algorithms. We have automated monitoring systems in place to detect errors that occur widely and bring them to the attention of humans, but it is not possible to catch and fix every possible case at the scale at which we operate.

You can use our contact form to contact us but we do not provide any guarantees of responding or fixing the problem within X business days.

Regarding the paper "Heat generation in plasmonic nanostructures: Influence of morphology", we have provided the citation data as it appears at the source.

<meta name="citation_firstpage" content="153109" />

I believe for some physics journals it is standard to cite the first page only instead of a page range.

Regarding the paper "Super-resolution in high-contrast media", it is our mistake that we have classified the venue as a conference instead of a journal. Thank you for bringing this issue to our attention.

5

Pont is right that Google Scholar has a contact form, which is mainly useful if bugs in their own system causes mistakes. I'd hesitate to call this a mistake, however. At least not on their part, as they presumably source their data from e.g. publishers and journals. Contacting the publisher directly may be more productive.

The reason for the pages error you mention is simply that, at least in physics, you usually only include the starting page in a citation. In a list of references, I'd expect those two papers to be listed like this (or similar):

Appl. Phys. Lett. 94, 153109 (2009)

Proc. R. Soc. A 471, 20140946 (2015)

By oversight the end page then sometimes gets left out from e.g. BibTeX files. If you download the BibTeX file from the first paper from the publisher's website or from ADS it again works this way:

pages = {153109},

which, of course, is enough information to produce the above citation. As for the second paper from Royal Society, the BibTex file doesn't even contain a page reference! Still, in the PDF version and their table of contents, they give the citation with just the one number. The ADS version actually gives a range, but it's just

pages = {20140946-20140946},

(Google Scholar classifying it as a proceeding might be on them though, as the two other sources correctly identify it as an article.)

If you're using a reference style that requires including the end page you have my condolences. Otherwise, for your own sanity, I recommend accepting that "pages" sometimes just means "page information needed to produce the citation". Different databases, different publishers, different journals just aren't fully consistent with each other - especially when it comes to citation data that isn't necessarily required.

  • 2
    You make a good point about the data being sourced from publishers / journals; presumably, at least some systematic errors of this kind arise from how the publisher is presenting the data (e.g. they may find that presenting it a different way helps Google pick it up better). It may be worth contacting the publisher themselves, as they may be able to deal with the error or pass it to whomever handles their data for them. – Myles Mar 14 '18 at 17:53
  • Yes, I think you're right about contacting the publisher - it certainly is easier to correct data at the source than later down the chain. – Anyon Mar 14 '18 at 18:20
  • Thank you for your output. I thought so at the beginning, but diving inside the papers themselves,that incredibly high number sometimes was constant along the paper, so it didn't gave me an impression of being a page number; Furthermore, I was suspicious of a 20 million page issue. Anyways, google scholar replied with a generic "deal with the Journals", so at least I tried. – pancho Mar 14 '18 at 21:12
  • Can't do more than try. Anyway, sometimes you get what seems like artificially high numbers, but it's still the number you're supposed to cite. For example, some journals use running page numbers that don't reset between issues. In the case of that 20 million number, I suspect the 2014 part is the year when they first received the manuscript. It would be more of an article identifier than a page number in that case, but it's still the way to find the article. – Anyon Mar 14 '18 at 21:25

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