13

I am an Computer Science undergraduate student and I am looking for my potential mentor for my PhD study. I guess besides research interest, personality etc., h-index is an important factor one should consider. So I wonder whether there is a way to find out about the precise h-index of an arbitral author.

I've been using Microsoft Academic Search. It's cool, it often finds the right person and gives you his/her h-index. However, I believe that this tool has underestimated the h-index of many researchers. (I think so because I've tried typing in some very famous professors' names and the h-index that Microsoft gives is about 10~20 lower than the actually value).

Moreover, since I am mostly interested in Human-Computer Interaction so I am also referring to ACM SIGCHI's Most Frequent Authors Page. The page is awesome: it gives you the most brilliant professors ranked by their number of publications. However, I think h-index is a better metric than the number of publications alone.

I could, of course, go to the professors' websites and read their publication list. But I want to know whether there is a better, more efficient way of finding h-index.

CherryQu
  • 1,525
  • 1
  • 10
  • 20
  • Google scholar is another possibility, but this is likely to be inflated. – Dave Clarke Dec 12 '12 at 12:43
  • 14
    I totally disagree with your assessment that h-index is an important factor one should consider when choosing a supervisor. That said, it is irrelevant to your question about how to find the information. – StrongBad Dec 12 '12 at 13:07
  • Okay :) I just put it there as part of my motivation for finding the h-index. Meanwhile, I would be glad to hear other guidelines about finding a good supervisor. – CherryQu Dec 12 '12 at 13:19
  • 2
    @CherryQu please ask a new question about how to find a good supervisor (have a search through past questions also). Also if you start your comment with an @ (as I did here), the person will get a notification. – StrongBad Dec 12 '12 at 13:45
  • 4
    @CherryQu Things as h-indexes are accumulative, so older guys have it a lot higher (don't even compare people of different age with h-index!). Whereas, looking at friends, it seems to be the best idea to get a very young, energetic prof, just starting her/his career. There is more energy, fresher topics, (s)he is still hands-on in research and can have time for you, on daily basis (also, because then usually group isn't that large). But still, there is huge variance with respect to personality, topics etc. But they get low h-indices, since they are not cited yet and there are just starting. – Piotr Migdal Dec 12 '12 at 13:57
  • excellent point @PiotrMigdal: as a young apprentice, would you go work in a 1-michelin star restaurant striving to get its second star, or would you go to a boring, posh, immaculate 3-michelin star where they serve the latest food craze? – ElCid Dec 12 '12 at 16:01
  • @DanielE.Shub: it seems to me that the "social web" at large is giving the opportunity to anyone to "say something about themselves" in a numeric format. This stackexchange is an example (how large is your reputation?), and the h-index is of course another example. Clearly there are ways to game these systems: on MetaStackOverflow there are dozens of questions about the issue. But these numbers are used nowadays, in more or less formal contexts: when I was interviewed for my new post, I was asked for my h-index, both to show that I knew it, and to place my research in a larger context – ElCid Dec 12 '12 at 16:16
  • @ElCid there very well might be times where the h-index is informative and important. I stand by my statement that it is not important for choosing an advisor. – StrongBad Dec 12 '12 at 18:47
  • 1
    Not a real answer, but relevant: look also here: http://academia.stackexchange.com/questions/4793/ – walkmanyi Dec 12 '12 at 20:32
  • 2
    when I was interviewed for my new post, I was asked for my h-index — Don't walk. Run. – JeffE Dec 13 '12 at 15:37
  • "h-indexes are accumulative, so older guys have it a lot higher (don't even compare people of different age with h-index!)" At least in some fields h/(years since first publication) isn't a bad metric for sustained output. Of course, it still does not do to compare across subdisciplines, so it is always better to compare someone's h to that of their peers rather than to random other acedemics. – dmckee --- ex-moderator kitten Dec 14 '12 at 20:24
  • h-index is almost useless as a measure of how good an academic is and just tells you how long they have been hanging around and publishing continuously. James Maynard only has h-index of 11 on Google Scholar and below 1000 citations, and he has been awarded the Fields Medal. – Tom Jul 09 '22 at 17:44

4 Answers4

11

There is no such thing as "the precise h-index".

The h-index is defined as the largest integer h such that at least h of the author's publications have at least h citations each. But what's a publication? Do first-author publications carry more weight? Do ArXiv, technical report, conference, and journal versions of the "same" paper (all of which may have citations) count as four papers or one? Or does it depend on the difference in content between versions? (How do you measure that difference?) Does the quality of the venue matter? If so, which venues count as "real" publications? Do survey articles count? Popular science articles? Blog posts? StackExchange questions?* Do self-citations count? If so, does a citation to a paper by X and Y, in a paper by Y and Z, count toward X's h-index? How much (and who) are you willing to pay to make sure you've really counted every citation to every publication?

The different sources of the h-index make different well-reasoned decisions about each of these issues, none of which make sense for ALL areas, even within computer science.

I think h-index is a better metric than the number of publications alone.

This is a pretty low standard. You can do better.

*At least one of my StackExchange questions has more citations than at least one of my papers.

JeffE
  • 98,599
  • 14
  • 236
  • 398
  • 1
    interesting questions Jeff.. How much (and who) are you willing to pay to make sure you've really counted every citation to every publication? That's the big question. Shocked by the SE citation; what is the objective of citing a SE question? – seteropere Dec 13 '12 at 05:20
  • 3
    what is the objective of citing a SE question? — The same as the objective of citing anything else: Giving the author credit for their ideas. Why does it matter whether the idea was formally published? – JeffE Dec 13 '12 at 05:49
  • I may say credibility (if the word correct). i'm new to SE but citing online discussions in scientific papers seems odd to me. – seteropere Dec 13 '12 at 06:03
  • 1
    I have cited a MathOverflow discussion in one of my preprints. I asked for the solution of a small problem there and I got an answer; this was the only way I had to give credit. – Federico Poloni Dec 13 '12 at 08:42
  • Frontiers in Computational Neuroscience is full of papers on "doing better" than the h-index in some sort of automated way. If you have clearly better ideas of how to do it, maybe you can share (or publish them, if you haven't already)? It's a tricky problem. – Rex Kerr Dec 13 '12 at 16:55
  • 2
    It's not a "tricky problem"; it's impossible even in principle to automate evaluation of research quality. – JeffE Dec 14 '12 at 01:59
  • 1
    @JeffE: although I agree that it's not possible to evaluate the absolute quality of research, it's being done already, by many institutions, that you like it or not... Any research project is evaluated on "quality", on various factors (the infamous "impact"...?) and on a scale between 1 to 5. Papers quality has started being evaluated on the number of citations; researchers will be evaluated on their h-indexes (or similar...). In my view it's an inevitable process, due also to the current collapse of the university system and traditional "approach" – ElCid Dec 14 '12 at 10:41
  • 1
    @ElCid: Of course. And for precisely that reason, I feel compelled to argue against the practice at every opportunity. (What "current collapse of the university system"??) – JeffE Dec 14 '12 at 14:47
  • @JeffE: maybe the analysis below is not 100% correct but it is anyway an interesting read – ElCid Dec 15 '12 at 08:55
8

the h-index is an interesting measure, but it changes all the times

  • microsoft search does it
  • Google author search does it (see an example here)
  • the freely available tool PoP does it

you could try and triangulate the numbers that you get and find the average, as a possible solution

ElCid
  • 3,865
  • 19
  • 30
3

h-index is not an important factor to consider potential supervisors rather recent projects and publications count more for me. Also, h-index might misguide you (it has its limitations).
For example, if a new professor published 10 articles each cited at least 200 times the h-index will be only 10 ! .. Also, old researchers have advantage over young ones.

Returning to the question:

So I wonder whether there is a way to find out about the precise h-index of an arbitral author

I am not aware of any engine/program gives a precise h-index. There is no precise h-index out there and finding a mechanism to find it is an interesting work..

seteropere
  • 13,788
  • 6
  • 50
  • 87
2

I believe for the UK's upcoming 2014 REF exercise the bibliometrics are based on the Scopus database (http://www.ref.ac.uk/background/bibliometrics/). This likely has to do with both Elsevier political power and the quality of the database.

StrongBad
  • 104,237
  • 30
  • 262
  • 484