Generally, it is difficult to define a generalized principle which conference/journals to choose from as it varies across disciplines. But there are some basic rules to rule out "just purely bad conferences/journals". If you submit your work at these places, you will first regret forever, second, you will learn literally nothing, and third your reputation will be greatly damaged, so avoid at all cost.
- For conferences, the basic number to look for is acceptance rate. Although acceptance rate varies across disciplines, usually any good conferences have a 10%-30% acceptance rate and have roughly more than 100 submissions. An acceptance rate of 20% but with 5 submissions also does not make sense. Do not submit to conferences with 80% or 90% (some even 100%, i.e., with no peer-review process), they are not useful for your future research, it can also damage your reputation. We can sometimes look at the publishers, conference papers published by ACM or IEEE are usually good but there are some exceptions, though. H-index is also a fairly reliable metric to compare the ratings between conferences.
- For journals, there is a lot of controversies here but the basic number here to look for is the impact factor. Impact Factor is a bit controversial as it greatly differs across disciplines, so the key principle as a good human is not to compare which field is more impactful using this parameter. H-index is becoming more used for rating journals so you need to keep in touch for that. For Computing an impact factor of 2+ is usually considered decent so you should aim for that. Avoid submitting to journals that have (1) no impact factor, (2) no peer-review process, or (3) has volume number of 1 (or 2) which indicates that it is fairly new; try submitting to legacy journals which are usually the top-tier and reliable journals in the field. For publishers, ACM, IEEE, Elsevier, Springer, Taylor-Francis are usually trusted by academia.
One good point of venues that support rigorous peer review process is that you will learn a lot from all the experts and gurus. Experts often serve as a reviewer in such venues thus it is a good place to see how far are you from the "world" knowledge. My experience is that one top-tier paper trumps over 100+ crappy papers, so focus on publishing only one really good paper as your first starting point as a researcher.
There are pros and cons between submitting your work on a journal or conference. Conferences also have a quick review cycle, taking only 3 months. However, conferences are not easy in the sense that you are not allowed to revise your work significantly and resubmit. On the other hand, most journals allow you typically two major revisions which allow you some breathing time (usually two months for each revision) to fix your work according to reviewer comments. Anyhow, journals have very slow review cycle, mostly take 1 year or more.
One good strategy is to first aim the best possible, i.e., the flagship conferences/journals. This is because you will learn the most through the process. People told me that if you aim for the sun, even you fall down, you still find the stars (kinda philosophical I know). That’s exactly true here. I have a strong belief that only researchers who continue working at the top-tier conference even did not get acceptance are still "genuine" researchers. Do not be ashamed if it gets rejected, all top researchers got many rejections even when they become seniors. If you know how it is not research. What distinguishes them is their courage, ego (sometimes it is good) and "never give up" attitude.
Learn everything from your reviewers. A good strategy is to resubmit to these venues next year or resubmit to more specialized venues, with the condition that you fully revise your work based on reviewer comments.
Our lab is still very young, but all students must strive to publish in top-tier venues. So start reading from here. We primarily work in DL/NLP, HCI, and Brain. Here are the top venues:
1. https://www.sciencedirect.com/journal/neurocomputing (Neurocomputing)
2. https://iopscience.iop.org/journal/1741-2552 (Journal of Neural Engineering)
Note: For deep learning, ICLR, NeurIPS, and ICML are the top three conferences.
Note2: In machine learning/deep learning, conferences are valued more than journals.
Note3: You can compare h-index of journals/conferences in google scholar, to know the ranking of conferences/journals. For example, for NLP: https://scholar.google.co.th/citations?view_op=top_venues&hl=en&vq=eng_computationallinguistics