20

I just started my PhD in computer science (first year) and I enjoy reading the high-rated journals related to my field. It is true that I find them, sometimes, complicated (in some sense) and I cannot continue the reading. The reason why I like reading such journals is that (1) there are many beautiful ideas, (2) solid works and (3) I would like to submit to these journals in the future.

I think, however, that this strategy is not quite good as a junior (student). May be I should read other conference papers and journals.

Can you please suggest some useful advice.

  • 6
    Conference papers contain more cutting edge material than journals in Computer Science. – Dave Clarke Jul 1 '15 at 20:21
  • 2
    it really depends on why are you are reading. to keep afloat of new developments in your field? then skimming through top journals is probably the way to go. to get a deep understanding of your particular research topic? then you cannot restrict yourself to just a few high-ranking journals. maybe the piece of information that you really need is hidden in some obscure technical report. it is impossible to answer your question until you clarify your motivation. – henning Jul 1 '15 at 20:47
28

I've never been one to sit down and read a journal edition cover to cover. I think you're better off skimming titles and abstracts for articles that pique your interest. You should do this for conference proceedings too. You can't attend every talk at a conference, nor should you really expect yourself to read every article.

At some point you will need to narrow your interest to something pretty specific in order to develop a dissertation. You will also need to develop depth and historical perspective on your chosen topic. You should probably spend as much, if not more, time reading interesting individual articles and all their important references (and so on!) as you should trying to read whole journal editions or entire conference proceedings. There's value in a depth-oriented approach to reading articles.

I find that there's a pruning strategy of sorts that keeps you out of stuff that's not interesting. Sometimes randomly selected articles can be worthwhile, but there's lots of stuff out there that will be boring to you. Using a good article as a guide and reading some of its references can be much more productive.

13

The gold standard is indeed these kind of journals. I always tell my students: you are what you read :) This is because we learn by osmosis. You need to learn what is 'solid work', how to do technical writing, the vocab, and the most important ideas in your field. The latter is critical. If a reviewer finds that you missed a seminal work, your paper will be rejected.

In CS there are top conferences too, so do read them as well. If you are in networking then SIGCOMM and MOBICOM have very nicely written papers and usually nice ideas.

It is important not to discount lower tier venues. They may contain interesting ideas that may spark a wild fire. In other words, read them for ideas and conduct your research as per the gold standard.

  • 3
    I disagree with the first sentence. The gokd standard in CS (if there is one) is in high-quality conferences. Journals are secondary. – JeffE Jul 2 '15 at 13:05
  • 3
    @JeffE Gold-standards for novelty, maybe, but certainly not for quality (meaning correctness and presentation). – Raphael Jul 2 '15 at 15:30
8

The value of reading high-ranking journals in your field is that it will give you a somewhat broad view of what leading researchers in your field consider to be "good" and "important" research.

However, this "breadth-first" approach will not help you very much in your own research, including getting your own papers published.

Instead, I suggest you pursue a "depth-first" approach where you find "seminal" papers in the sub-fields that interest you the most, and then read all the important papers that cite those seminal papers, especially papers that argue against the seminal/foundational papers. In this way, you learn about the scientific discourse involved in any particular line of research.

You can find seminal papers several ways. Sometimes, they are highly cited. Other times, they are the focus of special issues of journals or special conferences. Sometimes, they aren't highly cited (i.e. do not have a high number of citations themselves), but they are cited or used in a few critical papers which in turn led to a significant line of research.

With this sort of analysis, you will learn the skill of evaluating papers not in isolation but in the context of a whole line of research, including research by detractors.


I advocate the "depth-first" approach to support your own research because in your dissertation you will need to take a position through your thesis statement(s) on a few research question(s). For this purpose, it is of little value to know all the latest research in the best journals. Instead, you need to know all (or most) of the research in one or a couple of lines of research, and especially the gaps in that research which you intend to fill with your dissertation.

4

To be honest there are some serious issues with the really "high-level" journals (namely nature and science). It is seemingly the case that the editors of said journals have a designated "hot topic" which they find to be of interest. A symptom of this drive is that the reproducibility of the experiments described in such articles suffers, (as evinced by e.g. that guy from Bell Labs a few years ago who was doing some seemingly phenomenal stuff with magnets). With this said, the acceptance of a topic in such a journal is usually indicative of the problem being of great interest. Depending on your intended field I would narrow down your search to journals which are more myopic as due to the breadth covered by such high-level journals searches in smaller journals I find tend to be more fruitful (as in I have to parse fewer abstracts/titles for something of interest).

  • I think interpreting "high level" as "Nature or Science" is a mistake. It seems clear that the OP means top specialized journals in computer science. – Tobias Kildetoft Jul 2 '15 at 12:37
  • 1
    To be more direct: Nature and Science are not high-rated journals in computer science. – JeffE Jul 2 '15 at 13:03
  • The comments here point toward a broader issue. To find out which journals are the important ones in your field, you should not rely on citation counts or similar rankings. Rather, you should ask your adviser and other faculty in your area. Also, it is likely that you will not get information directly in the form "journal X is important"; more often, the information will be "you should read paper Y." After this has happened for a lot of Y's, you'll have a sense of the importance of journals, just by seeing where those Y's were published. – Andreas Blass Jul 3 '15 at 23:53
3

I found/find that reading 'high rated' journals is a good place to start when investigating a research topic. Not only do they provide a solid grounding in contemporary 'advancements' in your field but they also have the potential to provide a huge resource by way of the reference section. Perusing reference material to which the author/s refer and upon which they preface their research can [over a number of journals] provide a listing of relevant and seminal works that minimises possibly of missing something important.

I also agree with the above comments re: scanning abstracts/titles etc. but would also urge you to seek out journals by particular (reputable) authors/researchers who are in your field and who will more than likely have written journals that could provide some insight into how their research evolved.

2

Reading and educating yourself as a new PhD student is required to improve your skills to, eventually, write something similar to what you are reading now. If you feel that the level of the papers you read is too high and you cannot follow it, then it is a good starting point to define the areas where you need to improve (given that you are reading in THE related work to your topic). After defining these areas, it is your task now to start working on it by attending lectures offered by your university, MOOCs, reading books, etc. You can also participate in cs.stackexchange or similar platforms that are related to your research, start reading the related questions, asking questions, and even answering questions. You can test this with two or three papers that you like most, but you are not able to follow, then you can come back again to these papers and test how you improve. Starting PhD is always not easy in CS, do not lose your enthusiasm and try always to improve your self.

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.