|
I work outside of academia, but would still like to learn about cutting edge topics in NLP. Not attending conferences and not having access to a university library makes this a little tricky. Any recommendation for websites that post new research/conference/journal papers (freely or cheaply)? Blogs that review interesting new papers? Conferences I should read the publication lists of? I'm personally most interested in NLP, then machine learning more broadly. But if you've got recommendations for some other field relevant to MetaOptimize, that'd be worthwhile too. |
|
You probably want to stick with conferences because many journals charge money for their articles. If there is a particular article you are interested in though, you should check the authors' website because many authors make a copy of their article available there. Each journal has different rules for the rights retained by authors when they publish articles, so it's hard to know when they will be available. Here's a list of conferences I typically end up reading articles from that have strong NLP tracks: NAACL, ACL, EMNLP, NIPS, UAI, AISTATS, and ICML. =joe For NLP one can also check the ACL anthology, which is open access and keeps a sizeable share of the publications in the area.
(Feb 10 '11 at 14:28)
Alexandre Passos ♦
2
Second the part about checking author's websites. My university doesn't have access to some of the non-core journals, but the articles are usually available on the author's website. You can usually read the abstract easily enough online, so if something is really important you can also email the author.
(Feb 10 '11 at 17:29)
Robert Layton
I forgot that computer science journals typically allow professors to put a pdf of the article on their website. I am actually a psychology graduate student and probably around 1/2 of the relevant psychology journals allow you to put pdfs on your website. =joe
(Feb 10 '11 at 17:59)
Joseph Austerweil
|
|
I suggest CiteSeer: |
|
The Machine Learning Journal from MIT is a great resource, a bit on the theoretical side though. You can also try to look for papers in Google Scholar, most of them are freely available in the authors webpages. (Remember to click "See all versions" if one of them charges) I might recommend you to look into Stanford, MIT and Berkley's Machine Learning and NLP Labs, most of the time their paper are freely available there as well. |
|
Bing's academic search is very good, imho better than Google Scholar. 1
Tried it, I like UI and RSS instead of email alerts, but it's not a substitute for so much smaller index! Google Scholar finds about an order of magnitude more papers for me.
(Feb 11 '11 at 06:24)
ivank
|
|
CiteULike & Mendeley are nice complements to tools like CiteSeer & Google Scholar. They are basically social/collaborative sites centered arond academic references. They allow you to bookmark, organize, share & tag references and you can upload your personal copies if you have downloaded them from elsewhere like CiteSeer or Scholar. There is also a nice iPad client for Mendeley, so you can sync & read your papers on the go. Connotea is also in this category but I have never used it, I think it is specialized towards the bio sciences. Some of these sites also have recommendation features |
|
Personally, I absolutely despise all the vendors that serve up research papers. So far what has worked for me:
Generally I've found, if the paper is worth my time in the first place the individual who published it is probably proud enough of their work to want to get it out there instead of relying on services like CiteULike to serve it up to their audience. Some examples:
Clearly I have a bent toward DBN research, but I believe the same approach will work with other areas of ML. |
