Best journals for computer science?
July 21, 2024 6:19 PM Subscribe
Is there publicly accessible ranking of CompSci peer reviewed research journals and their relative importance?
My brief Google searches for "comp sci impact factor" brings up lists that don't mention USENIX at all, and I can't tell if this means the lists are incomplete, or if say OSDI just doesn't as rank highly as I imagined.
Normally origin doesn't matter too much when I come across a paper, but I'd like to start finding papers that I can apply professionally faster, and ideally soon after publication.
My brief Google searches for "comp sci impact factor" brings up lists that don't mention USENIX at all, and I can't tell if this means the lists are incomplete, or if say OSDI just doesn't as rank highly as I imagined.
Normally origin doesn't matter too much when I come across a paper, but I'd like to start finding papers that I can apply professionally faster, and ideally soon after publication.
Best answer: On failure preview: you actually linked a conference! And it's co-run with ACM! And seems like the sort of thing where I'd expect such work to be published. So I'd expect the proceedings to have an impact factor somewhere. Huh. Got me... Do note how my opinion did a 180 when I saw ACM was co-organizing the conference. Thus is the power of High Impact Factors In The World Of Academics and Academic Refugees. A quick google suggests that in 2003, OSGI was high impact but I can't find a contemporary version of that list.
posted by Alterscape at 6:40 PM on July 21, 2024
posted by Alterscape at 6:40 PM on July 21, 2024
Best answer: I've heard folks reference the CORE rankings for various things, and I'd trust those. (OSDI is in the highest rank there, fwiw, meaning it is a "flagship conference, a leading venue in a discipline area.") And this list looks reasonable at first glance for finding prominent conferences. Their methodology is based on citation counts, it looks like.
And yes, for most areas of CS, you want to be looking at conferences, not journals, especially if you want cutting edge work.
posted by whatnotever at 6:58 PM on July 21, 2024 [2 favorites]
And yes, for most areas of CS, you want to be looking at conferences, not journals, especially if you want cutting edge work.
posted by whatnotever at 6:58 PM on July 21, 2024 [2 favorites]
Best answer: Scimago does have a freely available journal ranking: SJR. But like others said it's only for journals, not conference proceedings, and that's where CS stuff gets published first.
posted by kbuxton at 8:46 PM on July 21, 2024 [1 favorite]
posted by kbuxton at 8:46 PM on July 21, 2024 [1 favorite]
Response by poster: IEEE and ACM are the organizations behind most of the high-impact conferences my peers and I published in.
I know of IEEE and ACM, but they seem to be running comparatively vast numbers of conferences -- IEEE claims to sponsor over two thousand annually! And ACM's HCI has like fifty tracks, which feels pretty absurd.
And IEEE in particular has problems with tortured phrases, which are often interpreted as symptoms of disguised plagiarism. So I'm looking to something a bit more informative than the top level brand considering they cover what feels like 80 percent of the total publication space.
CORE rankings for various things
So I searched for field of research 4606 ("Distributed computing and systems software") in that database and one in particular jumped out as A* but also had 50 comments entirely of 5.0 ratings, when most have 0 comments at all. Obviously I can rule out that sort of outlier, and I have no idea if or how that factors in ranking but Campbell's law immediately comes to mind.
And this list looks reasonable at first glance for finding prominent conferences. Their methodology is based on citation counts, it looks like.
I could swear I searched that list for usenix and came up blank. Well, this is why we seek independent confirmation ^_^
But like others said it's only for journals, not conference proceedings, and that's where CS stuff gets published first.
As a (former) CS student, I never really understood the difference. Is the idea that journals publish without a conference and lecture series attached?
posted by pwnguin at 9:01 PM on July 21, 2024
I know of IEEE and ACM, but they seem to be running comparatively vast numbers of conferences -- IEEE claims to sponsor over two thousand annually! And ACM's HCI has like fifty tracks, which feels pretty absurd.
And IEEE in particular has problems with tortured phrases, which are often interpreted as symptoms of disguised plagiarism. So I'm looking to something a bit more informative than the top level brand considering they cover what feels like 80 percent of the total publication space.
CORE rankings for various things
So I searched for field of research 4606 ("Distributed computing and systems software") in that database and one in particular jumped out as A* but also had 50 comments entirely of 5.0 ratings, when most have 0 comments at all. Obviously I can rule out that sort of outlier, and I have no idea if or how that factors in ranking but Campbell's law immediately comes to mind.
And this list looks reasonable at first glance for finding prominent conferences. Their methodology is based on citation counts, it looks like.
I could swear I searched that list for usenix and came up blank. Well, this is why we seek independent confirmation ^_^
But like others said it's only for journals, not conference proceedings, and that's where CS stuff gets published first.
As a (former) CS student, I never really understood the difference. Is the idea that journals publish without a conference and lecture series attached?
posted by pwnguin at 9:01 PM on July 21, 2024
As a (former) CS student, I never really understood the difference. Is the idea that journals publish without a conference and lecture series attached?
Conferences have talks and posters and workshops and panels and keynotes and so on, and they may publish "proceedings" with papers for some or all of those. Journals are just... collections of published papers. In a lot of fields, conferences have very low bars for entry, such as submitting an abstract of your talk only with no data, details, or other "meat." Then, they do not publish proceedings, or if they do, the published papers are not "worth" much because of the low bar for entry. In CS, conferences are often fully peer-reviewed, such that presenters submit complete papers for review in advance, acceptance rates can be very low, and the proceedings are then full of proper peer-reviewed scholarship.
In any field, journals also accept complete papers for review, accept a small number, and so on. It's just that in most fields, the journals are the main or only place for high-quality, peer-reviewed work, whereas in CS, conferences have that as well. Furthermore, in CS, the conferences in many subfields are seen as more important and impactful than the journals in those subfields. Often, big important work is published in a top conference first, and then later a longer paper about that work with more details may be published in a journal. The journal paper may be considered more "archival" in that case, and most people who cared were paying attention earlier to the conference paper. Very roughly, I'd look at conferences for new ideas and at journals for deeper explorations of older ideas.
posted by whatnotever at 9:32 PM on July 21, 2024 [4 favorites]
Conferences have talks and posters and workshops and panels and keynotes and so on, and they may publish "proceedings" with papers for some or all of those. Journals are just... collections of published papers. In a lot of fields, conferences have very low bars for entry, such as submitting an abstract of your talk only with no data, details, or other "meat." Then, they do not publish proceedings, or if they do, the published papers are not "worth" much because of the low bar for entry. In CS, conferences are often fully peer-reviewed, such that presenters submit complete papers for review in advance, acceptance rates can be very low, and the proceedings are then full of proper peer-reviewed scholarship.
In any field, journals also accept complete papers for review, accept a small number, and so on. It's just that in most fields, the journals are the main or only place for high-quality, peer-reviewed work, whereas in CS, conferences have that as well. Furthermore, in CS, the conferences in many subfields are seen as more important and impactful than the journals in those subfields. Often, big important work is published in a top conference first, and then later a longer paper about that work with more details may be published in a journal. The journal paper may be considered more "archival" in that case, and most people who cared were paying attention earlier to the conference paper. Very roughly, I'd look at conferences for new ideas and at journals for deeper explorations of older ideas.
posted by whatnotever at 9:32 PM on July 21, 2024 [4 favorites]
Journals publish longer, more fully fleshed-out work. My only published journal article (I'm forever ABD) is something like 20 pages long, and combines and extends work that was published in three or four conference papers over several years. Getting it published was about an 18-month flow after writing the first submitted draft of the bloody thing, from initial submission through a couple of rounds of rewrites and a significant additional study that the reviewers requested. Conference papers, on the other hand, are between 3 and 10 pages, include a smaller amount of material, and take somewhere between 3-6 months from submit to final acceptance, usually including at most one round of revisions. Also, most conferences are annual and broadly take the same kind of work every year (I spent many years after I dropped out working as research support staff in academic contexts, and our year was often organized around collecting data early enough to hit significant annual submission deadlines); journals tend to bundle related topics together so even though they publish more frequently, you might have to wait for an editor to decide your topic fits in a particular issue.
When I was going through the process, a "good" Ph.D student was supposed to publish on the order of 7-10-ish conference papers and 2-ish journal articles, and maybe a book chapter if you were really exemplary. So the flow was usually to get your ideas out there in conference papers, and keep building on the ideas until you had enough "meat" to justify a journal article, which wasn't really something anyone looked at (they were reading your conference papers, instead, because that's where your newest ideas were) but was necessary to keep The Powers That Be satisfied (and, on preview, yeah, archival/summary of a trajectory of work).
On the topic of the huge number of conferences, yeah, that sounds about right. Most fields I know about these days (computer graphics, HCI, UX, ML) have one big conference that is high impact, and a couple of smaller conferences that are either local and provide a venue for work where maybe the authors can't get funding to travel to the flagship conference (compare SIGGRAPH and EuroGraphics / SIGGRAPH Asia) or are just lower prestige (compare IEEE VR to.. whatever ACM has. I know they have something but nobody I've worked with cares about it). There are a lot of smaller conferences too. When I was a grad student (at a US R1 university with a chip on its collective shoulder), we never even targeted the small conferences, they were seen as low prestige and not worth the trouble.
posted by Alterscape at 9:38 PM on July 21, 2024 [1 favorite]
When I was going through the process, a "good" Ph.D student was supposed to publish on the order of 7-10-ish conference papers and 2-ish journal articles, and maybe a book chapter if you were really exemplary. So the flow was usually to get your ideas out there in conference papers, and keep building on the ideas until you had enough "meat" to justify a journal article, which wasn't really something anyone looked at (they were reading your conference papers, instead, because that's where your newest ideas were) but was necessary to keep The Powers That Be satisfied (and, on preview, yeah, archival/summary of a trajectory of work).
On the topic of the huge number of conferences, yeah, that sounds about right. Most fields I know about these days (computer graphics, HCI, UX, ML) have one big conference that is high impact, and a couple of smaller conferences that are either local and provide a venue for work where maybe the authors can't get funding to travel to the flagship conference (compare SIGGRAPH and EuroGraphics / SIGGRAPH Asia) or are just lower prestige (compare IEEE VR to.. whatever ACM has. I know they have something but nobody I've worked with cares about it). There are a lot of smaller conferences too. When I was a grad student (at a US R1 university with a chip on its collective shoulder), we never even targeted the small conferences, they were seen as low prestige and not worth the trouble.
posted by Alterscape at 9:38 PM on July 21, 2024 [1 favorite]
Also, you mentioned ACM CHI (but called it HCI, so I didn't notice until I re-read). I've presented posters and papers and attended workshops at like 4 of those and they're huge and bananas but absolutely a real thing. CHI is, or at least was when I was active, absolutely the hot-shit conference in HCI, and getting papers published there was really important if you wanted an academic career in that field. Like, in grad school, the #1 goal my advisor gave me was to have something there every year.
posted by Alterscape at 9:45 PM on July 21, 2024
posted by Alterscape at 9:45 PM on July 21, 2024
Response by poster: Also, you mentioned ACM CHI (but called it HCI, so I didn't notice until I re-read).
Apologies. But really that name invites the error, IMO.
And I didn't intend to imply it was fake in any way, just really, really big. Jeff Huang's best papers db for 2023 puts them at ~35 best papers to SIGGRAPH's ~5, so there must be a lot of publications!
posted by pwnguin at 11:55 PM on July 21, 2024 [1 favorite]
Apologies. But really that name invites the error, IMO.
And I didn't intend to imply it was fake in any way, just really, really big. Jeff Huang's best papers db for 2023 puts them at ~35 best papers to SIGGRAPH's ~5, so there must be a lot of publications!
posted by pwnguin at 11:55 PM on July 21, 2024 [1 favorite]
Yeah, the name is weird, and it is a behemoth. Most academic conferences are like, a nice hotel's conference facilities. CHI books entire convention centers, and fills them. Some of it is discipline culture tho (agree that SIGGRAPH is about the same size, and I've been to both a SIGGRAPH and a CHI in Vancouver and they both booked one wing of the conference center).
I apologize for being prickly. Even though I've been done with academia as an aspiring Ph.D for over a decade, still got some strong feelings rattling around in here (grad school does that).
Anyhow, you've got at least a couple of people with relevant background paying attention to the thread. If you have a more specific question re: publication venues we can probably help with answers or research strategies to get answers, so go ahead!
posted by Alterscape at 7:06 AM on July 22, 2024
I apologize for being prickly. Even though I've been done with academia as an aspiring Ph.D for over a decade, still got some strong feelings rattling around in here (grad school does that).
Anyhow, you've got at least a couple of people with relevant background paying attention to the thread. If you have a more specific question re: publication venues we can probably help with answers or research strategies to get answers, so go ahead!
posted by Alterscape at 7:06 AM on July 22, 2024
Best answer: Google Scholar publishes yearly rankings based on citations. Here are the general CS rankings, with subcategories available at the top.
*Usual disclaimers about all journal metrics being flawed in one way or another, but that doesn't mean they're not useful in certain circumstances.
posted by chrisamiller at 7:07 AM on July 24, 2024 [1 favorite]
*Usual disclaimers about all journal metrics being flawed in one way or another, but that doesn't mean they're not useful in certain circumstances.
posted by chrisamiller at 7:07 AM on July 24, 2024 [1 favorite]
« Older Seattle coffee for picky coffee snobs? | Where is the best place to sit for a baroque... Newer »
You are not logged in, either login or create an account to post comments
posted by Alterscape at 6:35 PM on July 21, 2024 [3 favorites]