Wednesday, December 14, 2011

Google Scholar versus Web of Science

Web of Science (WoS) is the de facto standard for citation information. It's citation counts are used for many purposes, among which to decide I am a good scientist. Web of Science, however, really expensive, and Joe the Plumber does not have access. No wonder, he doesn't know which scientist to trust (...).

Recently, Google made their Scholar product open to all, allowing you to list your publications (about my list), which Google with augment with citation counts. If you search the web, you'll find much being said about the two, in particular compared with each other. One aspect is the accurateness of the citation counts, as people are afraid gaming, and random noise found on the web. Others would (counter)argue that Google captures a wider range of literature.

So, I was wondering how this would reflect on my impact. I know that WoS is not errorless either, and I have been making various support requests over the years (my WoS records still have errors). So, do they complement overlap? Are citation counts comparable. In fact, this turns out to be true:

I would be drooling if I got this kind of regression in my nanoQSAR studies! :) There is a very strong regression, indeed. One of the advantages of Google Scholar is does not select an elite group of journals (of course, they have to, because there data analysis process involves much more human curation), while Scholar captures newer Open Access journals, like the J. Cheminformatics, too. While I may be a bit of a non-typical scientist (some even argue I am not even doing science...), the overall outcome is that Google Scholar is actually more accurate about my impact than Web of Science is right now.


  1. Hi Egon,
    I think one bad issue with Google Scholar is that self-citations are not excluded.

    WOS allows the exclusion of self-citations "View without self-citations" however fails in generating the statistics and h-index without self-citations.

    But i think its important to see those researchers that *extremely* pump-up their citations.

    And I think Scopus has a even nicer function to exclude citations from all former co-authors, basically showing really independent citations.

    One thing were WOS failed was the inclusion of the ResearcherID, which is a really nice tool. Once you are in WOS the author finder does not really work (try a generic name).

    WOS also has more conventional data such as conference abstracts (ACS) and Google has greater coverage of web-based publications, but not many (ACS) conferences.

    Anyway the real transformation for *all* sciences is coming from Google Scholar and Google Books and Google itself. For that alone Google should be awarded a combined Nobel prize of all sciences.


  2. Tobias, the point about self-citation is an interesting one! Have you asked Google to add that feature?

  3. Egon,

    What is the meta data (and units) for the two axes of your plot?

  4. The axes are the number of citations found for Web of Science (x) and Google Scholar (y).

    Besides the strong correlation (as compared to what I am used to see), there is a non-linear behavior too. I am not sure about the significance, but it may be explained by Google Scholar finding citations much faster than Web of Science, where it can takes months. Therefore, over the time of months, the citation counts would become more consistent.

  5. My correlations are even higher than yours (cannot attach the figure here) Just have a scaling factor of about 1.9 (i.e. Google citations are 1.9 more important than WOS). Correlation between WOS and Scopus is even more important. The worse is Scopus Vs Google (But still enormously significant).

  6. Egon,
    aehm, just did it. Also clicking through co-authors that are registered would be helpful.

  7. "Also clicking through co-authors that are registered would be helpful."

    Tobias, what do you mean with that? You can already create a list of co-authors, and then click through them...

  8. Egon,
    Google Citations misses links for co-authors.

    A) If you click on a publication title it will link to the DOI or PDF. What about the authors? No link. Even for authors that have a Google Citations ID.

    B) Not all co-authors are automatically shown in the right box.

    Check Yongquan Han, not in A) and not in B)


    1. Yes, that would be nice. I have filed a 'feedback' report with Google Scholar, point to your comment here.