12

I'm in computer sciences, in applied Machine Learning. I often have to switch applications and I find it cumbersome trying to discern the really good and bad works by doing a large survey.

Whenever you are new in a field, how would you direct your survey to find the people doing seminal work in the area, for example in the specific case of Latent Dirichlet Allocation in Computer Science, you can always go to a 2003 paper by Blei, Jordan and Ng. While if you are a seasoned researcher of the area, you already know this paper is relevant, how a newcomer would find such paper?

Do you ask experts in the area? Do you go by number of citations (I often find this deceptive, particularly in Bioinformatics)

I'm interested in this because I'm switching applications now, and I would like to know which group's work I should be following or which person's papers I should be reading.

|improve this question
  • 5
    Leon, I'm afraid the question as currently worded is awfully subjective. The “people doing most of the interesting work” will just not be the same depending whom you ask: what's interesting to you might not be to me, what's groundbreaking to you might be purely conjectural to me, etc. If you're asking “what's a good metric of ‘interesting work’”, I don't think it fits the format. Maybe you can reword it to ask something closer to “what are metrics other than h-index to measure ‘influence’ in the academic world?”, which fits the site better (I think). – F'x Nov 6 '12 at 22:48
  • 5
    Voted to close as not constructive. – JeffE Nov 7 '12 at 3:05
  • The question is IMHO relevant and I see its point, though needs to be reformulated. I vote for not closing this question. Will try to answer as well. – walkmanyi Nov 7 '12 at 7:52
  • In my opinion an answer that "especially in such cases the "leaderships" is subjective" is an objective answer. I vote to reopen. Not for everyone it needs to be obvious. – Piotr Migdal Nov 7 '12 at 15:30
  • 1
    @PiotrMigdal it's not an answer to the question asked (how). The question is subjective, and we don't judge the answers, but the question :) – F'x Nov 7 '12 at 15:55
10

Problem is that you hamstrung yourself by eliminating the only correct strategy: doing a long literature survey yourself. You wish to rely on proxies, and sometimes surveys can do that for you, but other people's opinions of "good" and "bad" are just that: opinions.

The only way to really get a sense of good and bad is to read a LOT, think a lot about what you read, discuss with people who might know more about the area, think a lot more, read more, and repeat.

|improve this answer
  • 2
    I've learnt than relying only on doing long literature surveys by myself was a bad idea (and I've learnt it the hard was; a few times). The reason is that you can easily miss a pearl just because it has not key works you are looking for, or is older, or is not cited enough (there is no way to lookup most citations at more that 1 level). – Piotr Migdal Nov 12 '12 at 20:19
  • 2
    very true: hence the "discuss with people" part :). I do that even now at conferences to get a sense of what's going on in different areas that I can't monitor myself. – Suresh Nov 12 '12 at 20:55
7

Well, one slightly unorthodox method for finding these might be to find the "society" representing the field, and find who their recent award winners are—particularly if they have "young investigator" awards that will recognize work considered to be of significant impact.

The relative advantage of this approach is that it it represents the consensus of people working in a given field; the disadvantage is that the consensus may be of only a handful of members on a committee. But these are people who in principle should know the ins and outs of their field better than most.

|improve this answer
  • I don't understand the downvote. Could the downvoter please comment? – Joel Reyes Noche Nov 7 '12 at 0:51
  • I guess the answer was too unorthodox. . . . – aeismail Nov 7 '12 at 8:07
6

My personal technique is to read the introduction to a number of research papers and find the most commonly cited papers/authors/author groupings. This will tell you who is the most "influential", i.e. whose work is driving the field and causing others to perform research branching of their original findings. This is similar to examining publication count, but more useful in that you can see how broadly the citations are applied, and judge from context whether the citation has merit.

|improve this answer
1

IMHO, the research work should be around good publications not good "people" even if they are highly cited. In almost any field you will find introductory papers and surveys discussing the problems and the current state of the art. Those are the beginning. If you find the text referring to something interesting to you; visit the references mentioned in that section/subsection/paragraph. Also, well-known journals usually have special issue for new/hot research areas.

|improve this answer
-1

Practical advise is to use platforms such as "ScienceDirect" or "Web of Knowledge". If you have institutional access, you can read every journal of a selected field. By the time you will read the same names repeatedly. Also helpful is to look through conference-papers and other documents prepared for conferences.

|improve this answer

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.