# Where to find published research papers?

Coming from the POV of someone who is thinking about pursuing a PhD in Computer science.

I am having trouble deciding what I would focus my research on when I go for my PhD. See also this question on academia.SE.

So I am thinking that reading/keeping current on what research is being done and what research papers are being released is a good source of....inspiration? Plus good knowledge to know.

Is there central place/database of/good starting place to see recently published CS research papers, or are they all hidden deep within the websites of the University they were written at?

-
Look for journals and conferences in the field you are interested in. –  Dave Clarke Dec 3 '12 at 22:03
Don't just read papers. Talk to a human being. Or failing that, a professor. –  JeffE Dec 6 '12 at 5:05
Read what you think is fun. And then read it with a critical eye. Can the result be improved? Can it be generalized? Can it be understood differently? How does it relate to X? Maybe the critical point is that you must be sufficiently interested to sort of ask these questions without even thinking about it. There are topics you grok and topics you do not. That matters, and only you can tell. The currently most popular topics are not necessarily the best ones. But once you are interested in something, you must check the litterature, new and old, for ideas, and to make sure you do original work. –  babou Jul 5 '14 at 10:33

As AJed mentioned, ACM digital library and IEEE xplorer would be near the top of list.

Additionally Goggling with advanced search for the name of the reference with the option to search for PDF or PS increases the change of a hit.

Sometimes limiting the search to cs.xyz or xyz.edu increases the quality of results.

If you are just starting out on a subject, try adding the word "survey" as a search term because many researchers write survey papers so that someone new to the subject has an informed person guiding the way.

I also like to hit CitSeerX, WorldCat and Google Scholar

Don't forget to look at the patents in Google Scholar, I don't usually but every now and then companies have to give out something good they may not publish anywhere else.

If you don't attend school, you can see if a school provides access to their library for outside third parties. I found out that I can do this for Princeton University for $300 a year. Added 3/2/2013 Microsoft Academic Search This can also visualize publication trends in computer - So, i can access any digital library Princeton is registered to with 300$/year only ?. That would be cool ! –  AJed Dec 4 '12 at 3:06
I only inquired. See General Access Information. They were quite nice when I called them a few months ago. If you leave near Washington D.C. and are a US citizen, you can use Library of Congress for free. It's great!!! –  Guy Coder Dec 4 '12 at 14:25
+1 for Google Scholar. –  Raphael Dec 4 '12 at 16:54
While not helpful for published research papers but helpful for finding info when starting to learn; add the keyword "notes" when searching. –  Guy Coder Jul 11 '14 at 11:47

See the ACM digital library, IEEE xplorer. These are the top in my opinion. Look as well in ScienceDirect (Elsevier) and Springer (for theoretical computer science, I believe these two libraries are better).

Usually, googling your research problem would lead you to papers. The journals in which these papers are published is what you are looking. Of course, use the references and citations of the paper you read. In the long term, you will restrict yourself with the journals of your field.

-

A very large fraction of new computer science papers come up on arxiv: http://arxiv.org/corr/home.

If you are interested in complexity theory give ECCC a look: http://eccc.hpi-web.de/.

If you are interested in crypto, there is the IACR eprint archive: http://eprint.iacr.org/.

All of these are free and frankly should suffice for most of recent papers.

EDIT: Maybe I was being too positive there. I mostly know about US-type theoretical computer science.

-
"A very large fraction of new computer science papers come up on arxiv:" -- that may depend on the subfield. Sadly, free resources are far from sufficient in many subfields. :/ –  Raphael Dec 4 '12 at 16:54
ok, so i guess i only know about data structures and algorithms, and complexity. curious though, which subfields don't post on arxiv? and i wonder why... –  Sasho Nikolov Dec 4 '12 at 17:09
I wouldn't know; I only remember the computer graphics and AI (?) people making a noise when our library cancelled the IEEE subscription. –  Raphael Dec 4 '12 at 23:14
well then i am not sure you have a point..for example the proceedings of AAAI, NIPS, ICML, and UAI are all online. of course no one wants subscriptions canceled, but it still seems to me that as a community we tend to make most of our (recent) work freely available –  Sasho Nikolov Dec 5 '12 at 0:28
@SashoNikolov, I have been informed that a lot of programming language types don't use arXiv (by an active, well regarded programming language guy), but he wasn't sure why other than cultural inertia. Being a theory person, I was puzzled too. –  Luke Mathieson Jan 13 '14 at 9:40

Microsoft Academic search is very good. You can select a list of conferences in your area and get RSS feeds of the latest papers.

-

If I'm looking for a particular topic or research problem, I usually start with Google scholar. To search for recent work by a particular author, title, or conference, sometimes DBLP is the way to go.

If you just want to browse recent work, then as others have said sometimes arxiv.org is a good way to go, but take care that the authors are reputable because arxiv is not peer reviewed. Another even better way to go is to follow what's been recently published in conferences in your field. I like confSearch for finding upcoming or recent conferences. You can usually find a list of accepted papers from a conference even before those papers are printed in the proceedings, and then locate the papers on authors Web pages or on arxiv.

-

In the other answers good sources are given. What I like to add is that not only you should try to locate single papers in a field, but also to find in what conferences they have been published. Then you can take some volumes of recent years and browse a little what has been happening there. Conference series might be from ACM, IEEE, or Springer LNCS.

Also, Google is a good help, but single papers in the arxiv do not tell about the "status" of the research. Research must have been accepted/presented elsewhere in order to count.

-
Nothing (except possibly time, and yet ...) will really tell you about the status of research. There are erroneous papers published in reputable venues of media, and there are very good papers they remain for a very long time (or forever) as tech reports from some company or institution. Overvaluation of the medium as a measure of research quality is one of the problems of research management. It does give some indication, but that is all. –  babou Jul 5 '14 at 10:12

In my opinion, since as you said you haven't decided yet the field of your research,starting from papers for inspiration is a bit odd. Firstly, if I were you I would search the Web for the general idea of "the state of the art" for each sector of Computer Science. Then you should go deeper reading papers, provided that you will have decided which path you would like to take.
Specifically now on which is the best way to browse a paper online, you should check IEEExplore, CiteSeerX (as the others mentioned) and CoRR and then if you can't find it for free, google the author and/or title to check if it is available on a non-charging repository. Good Luck!

-