Watch Out!
Student Media
Body of IITR
About Guide Get Involved

Career

Guest Editorial: All About Paper Publication

June 16, 2019
- Ekdeep Singh Lubana

I am writing this article to address certain myths that I have every so often been presented with while being consulted by peers who were interested in publishing their research work. I have had the opportunity to present my own work at three top conferences now, within the field of embedded systems, data compression, and sensors, while having published a journal paper too. Please note that this article is bound to be heavily biased by my own personal experiences, for academic communities have highly variable publication routines. For ex., while conferences garner a lot of reputation in circuit branches (possibly more so than journals), this may not be the case in natural sciences, which highly regard journal publications.

So, to begin with, there are several cycles involved in publishing one’s work.

  1. Building a manuscript–Your article/research paper is supposed to be a well-drafted manuscript that, succinctly but descriptively, demonstrates the problem statement studied, the solution proposed, and the contributions made. This manuscript is to be submitted for review, thereafter.

  2. Choosing a platform to publish–Broadly, one can publish via three platforms: (a) preprints; (b) conferences; (c) journals. I give brief descriptions about the three in this paragraph, but elucidate further details on choosing the exact venue in a later part of the article. Preprints are not technically publications, but an early body of work that you find promising. The idea here is to either mark your problem territory or unveil solutions/problems that require attention from the community. ArXiV is the most popular preprint platform and most certainly an academic’s newsboard. Conferences are generally the platform of choice if your work has gained a certain level of maturity, such as if you define a problem and in part solve it. Finally, the old school, high-prestige publications are journal papers, which usually carry a lot of lustre with them. These papers are heavily scrutinized by several reviewers before publication.

  3. Peer review–In most conferences/journals, one’s submission is assigned 3-4 reviewers who assess the novelty, plausibility, and clarity of your work, thereby scoring it on a predefined scale. These reviewers strive to find achievements and loopholes within your research, which you may be given a chance to address. I say “may”, for reviews can lead to either direct publications or rebuttals. In the former situation, reviewer comments are unveiled to you and, if the review score is high, your paper is accepted for publication directly. It is now at your discretion to modify your paper as per reviewer comments. For ex., say a reviewer said the caption of figure 2 is unclear and is vital to understanding the paper. In this case, it would be a good idea to change the caption. However, note that one is not allowed to perform significant changes to the paper. For ex., if you have found an optimization technique to improve your algorithm’s speedup from 2x to 3x, you are not allowed to modify your paper towards the same. The idea here is that your proposition may be fallacious, which needs to be reviewed, but since the paper has already been accepted, it cannot be reviewed again. In case significant changes are found, the paper is rejected.

  4. Rebuttals (the latter situation): The pipeline towards acceptance decision varies from venue to venue. In most top conferences and journals, however, one is given a chance to address the limitations found by reviewers. Certain venues allow for one to update the paper to address the reviewer comments, while others ask for limited text (<500 words, typically) responses. The idea here is to convince why the reviewer why your assumptions are correct or why the supposed limitation may not actually cover your target domain. Also, if you do think that the reviewer is right with his analysis, save some electrons and just thank the reviewers. :3

Finally, the review/rebuttal process culminates in the publication of your work. Now, having described the general idea, let me talk about certain specifics that I find important for people to know.

  1. Faculty involvement–Something that simply confounds me is the opinion that one needs a professor to be affiliated with one’s paper in order to publish his/her work. This is an absurd notion. You can publish alone, with your friends, with your grandparents, with your friend’s grandparents–whosoever you want, truly. In general, however, the authors of a piece of work will be the people who have contributed to it in some way. If the contribution is not significant, then such people are conventionally mentioned in the acknowledgement.

  2. The lust of journal publications–the publication cycle for journals is excruciatingly long, to the point where certain journals even take a year or two to go from review to publication stage. I am personally a heretic when it comes to journal publications, for these long winded time windows can actually become the bottleneck of scientific progress.

  3. Publishing half-cooked ideas to conferences–The following especially applies to circuit branches. This maligned notion for conferences has become a rather old-school opinion. In actuality, the competitiveness of conferences has risen to such high degrees that unless you have significant contributions to unveil, publishing in conferences is really hard. Further, the amount of content expected to be published in a top international conference paper in certain fields is several folds more exhaustive than that of a journal. Such communities have essentially accepted the limitations of journal publications and are progressing towards more dynamic and fast-paced publication cycles. One could argue that this adulterates the content quality, but there’s enough empirical proof that indicates otherwise.

  4. Do not publish for the sake of it: This point cannot be stressed enough. Students interested in pursuing academic careers are usually hounding towards getting themselves published; however, a publication at an unrecognized platform can be worse than no publication. Strive for top venues and settle for suboptimal ones if you have to, but do not publish for the sake of it.

All this said and done, an important point that I haven’t yet addressed is how to choose a platform to publish. Say you want to publish your work at a conference, then a good idea would be to find high H5-index conferences within your field at Google Scholar Metrics. For ex., a search for top conferences in computer vision at Scholar Metrics results in the following link:

https://scholar.google.co.in/citations?view_op=top_venues&hl=en&vq=eng_computervisionpatternrecognition.

You can find a relevant category on the website to find a suitable venue for your work. H5-index describes how many papers have had at least as many citations, in the last 5 years, at that venue. This isn’t an ideal metric to test a conference’s competitiveness, but certainly a metric that correlates well. Journals are usually evaluated by impact factors, than H5-index. I wouldn’t think too much about the metrics, frankly. The idea is simply to find a list of top venues. Thereafter, you go through some publications from previous years at that venue and try to determine if your work fits in well. Remember, the idea of publication is not just to unveil your contributions, but also to take feedback to improve your work. Lastly, some pointers specifically on conferences. While journal publications are carried out through the comfort of one’s home/workplace, conferences demand traveling to a venue and presenting your work. Presentations can either be oral or poster. Oral presentations are usually much fewer than poster presentations and typically involve a 15-20 minute talk by one of the authors, followed by a 2-5 minute questioning round. Don’t worry if you feel too shy to ask questions to the presenter amongst a hall filled with several luminaires from your field (evil wink), for presentation sessions are usually followed by coffee breaks and allow enough time for one to interact with authors of the papers that one finds interesting. But, there’s also a lot of really good free food available during these breaks, so do make sure you really have something worthwhile to give up all of that for. That said, your presentation really, really needs to be remarkable. Your presentation is you getting a chance to pitch your work to investors, who may choose to read your paper, take up your work, build upon/around it, cite you, make you popular, and get you a Nobel/Turing/Abel/other field-specific counterparts of the darn medal.

Now, due to societal obligations, I am expected to conclude this article, but, frankly, I do not have a conclusion. Therefore, peace out. :)