Comments on: A simple plea to all computer scientists

By: Kurt

Kurt — Fri, 17 Aug 2007 14:55:31 +0000

Laura,Thanks for the comment. Your citation influence work looks very interesting (from the point of view of a potential user of the system). I haven’t had a chance to download the viewer software yet, but I’ll be sure to do that when I have time. I hope you have success with the open source development, because it would be great to have this publicly available with a web-based front end on it.Of course, once the important pieces of research are identified, it would still be nice to have a lower price-point for purchasing articles…-Kurt

By: Laura Dietz

Laura Dietz — Thu, 16 Aug 2007 14:21:37 +0000

Kurt,I definitely agree with you that $30 is too much when you are not 100% sure the paper is of great value.I also see the point that there are publishers that need to make money. And AFAIK their terms do not permit the authors to put their PDFs online (although many researchers do it illegally).Of course we all hope that soon all publications are accessible for free. Until then I propose a different approach consisting of two ingredients.At ICML07 I published a paper about “Unsupervised Prediction of Citation Influences” (you can also try out the prototype here: http://www.mpi-inf.mpg.de/~dietz/citinf.html ). Given some seed publications it depicts the citation graph of papers that are strongly related, filtering out many citations / citings that are not really relevant. The second ingredient is that publishers (i.e. Nature) are thinking about the OTMI interface. This is an interface that basically gives access to the word histogram of the papers, which is fine for many machine learning methods (such as the Citation Influence Model), but does not reveal the human readable version, with which they earn money.Cheers,– Laura

By: Kurt

Kurt — Thu, 26 Jul 2007 02:46:56 +0000

Thanks for the pointer. I actually had a link to Papadimitriou’s Reading the Classics course in an old blog entry, but I didn’t think to check there.The funny thing about my search for the Karp paper was that I was over-specifying. By entering the full title, I was getting page after page of hits from other papers’ citation lists. On his web page, Papadimitriou simply referred to it as “Karp’s 1972 paper”, and sure enough if I do a Google search on “karp paper 1972″, it shows up half-way down the first page of results.

By: 知北遊

知北遊 — Wed, 25 Jul 2007 10:28:42 +0000

Christos H. Papadimitriou had given a course called “CS294: Reading the Classics”.http://www.cs.berkeley.edu/~christos/classics/cs298.htmlThere are e-prints of some classic papers, including “Reducibility among combinatorial problems.”I have a part-time job in university library. My job is about Document Delivery Service. Some great papers are requested copies very often, but, for academic publisher, it might be not worthy to scan old papers published decades ago.Besides, the frequent scanning or copying is really hurtful to the paper-print. I am convinced that digitalizing those classic papers is good for academia.