4

Recently Microsoft has released a dataset, Microsoft Academic Graph. I have used DBLP dataset before. I want to know what is the difference between these two datasets.

2

I am not very familiar with both data sets. But it is pretty clear to me that the Microsoft Academic Graph (MAG) data set is much more comprehensive than the DBLP one in at least one aspect: field of study. As far as I know, DBLP data set covers only computer science subject domain, whereas MAG's coverage goes far beyond that. For more details, see this relevant MSR paper.

  • @AyushiDalmia: You're quite welcome (feel free to upvote and/or accept :-). – Aleksandr Blekh Oct 2 '15 at 10:18
  • 2
    Aleksandr is correct, dblp only covers computer science. On the other hand, dblp is able to curate its data manually by human editors while Microsoft needs to rely on algorithmic entity disambiguation, (Although, to the best of my knowledge, some results from dblp are regularly imported to Microsoft Academic Search, too.) So the choice you have here is essentially either using a smaller, focused data set with a somewhat higher accuracy in identifying its entities and in correcting errors (dblp), or using a much larger, more expressive data set with a slightly lesser accuracy (Microsoft), – MRA Oct 3 '15 at 9:46
  • 2
    @MRA: Thank you for your comment. While it is targeting the OP, I read it with interest and I agree with your points. Just FYI: when you want somebody to be notified of your comment, either comment directly to their post, or tag them by name (using at-name format, as above), if you comment to somebody else's post (the post owner will always be notified). – Aleksandr Blekh Oct 3 '15 at 10:12
0

Microsoft academic search dataset is part of the initiative by Microsoft for providing searchable interface to users for searching through the corpus of research data in all areas of research. This includes publication, year, author or any such relevant information pertaining to the dataset. Whereas, DBLP dataset as an academic initiative from the university of Trier to create a dataset regarding the research publication in area of Computer Science. Therefore, scale and scope of both the dataset is different.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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