EDIT3: CVJC1 finished; you can read the transcript here.

EDIT2: I have still no response from the author, yet I think it should go as planned: 12/17/2010, 16:00UTC, we will discuss text mining in general basing on the article proposed by @onestop:

Jeffrey L. Solka. Text data mining: Theory and methods. Statistical Surveys 2008

and possibly also text mining tools in R, after the paper proposed by @chl:

Ingo Feinerer, Kurt Hornik, David Meyer. Text Mining Infrastructure in R. JSS 2008

If you want to come, please register to the event here.
(This is not obligatory or necessary to participate, but you'll get an e-mail notification)

See also previous question.

Based on the discussion on the chat yesterday (thanks for chl, csgillespie, Gavin Simpson and Shane), the initial idea for this clarified. So the CVJC:

  • will be held monthly probably Fridays, maybe 16:00UTC (this may be changing so that each major time zone will have equal chance to be in a comfortable time)
  • the discussed article(s)
    • will be picked on a meta thread like this before the chat event
    • must be OpenAccess or preprints/reprints, so everyone could read them
    • we (mods) will try to contact the authors and invite them to join the discussion (their acceptance will be a required constrain to use a preprint)
  • may also appear in a form of R package promotion event when package author shows a guided demo of its capabilities (this needs some further refinement of technical details)

So please, propose the articles as answers to this question. In a first approximation the first CVJC is scheduled for 16:00UTC 17.12.2010.

EDIT: To settle timetable, let's say the deadline for paper suggestions is 23:59UTC 14.12.2010.


2 Answers 2


How about this article from the same journal I suggested before?

Jeffrey L. Solka. Text data mining: Theory and methods. Statistical Surveys 2008; 2: 94-112. DOI: 10.1214/07-SS016

Reasons: open access, not very long (c. 12 pages of main text), not very mathematical (5 equations, no calculus or measure theory), seems quite a hot topic at the moment, I know nothing about it myself (so i'm not offering to 'present' it! In fact i doubt i can make the meeting this friday, sorry...)

  • $\begingroup$ @onestop And nicely complemented by a paper in the JSS about TM, Text Mining Infrastructure in R. I'm afraid I may not be available next Friday, but it's already a good start. $\endgroup$
    – chl
    Dec 13, 2010 at 14:06
  • $\begingroup$ Ok, this seems a pretty nice idea. @chl The time and date may change. $\endgroup$
    – user88
    Dec 13, 2010 at 18:25
  • $\begingroup$ Nice paper, but you should notice that while the paper is not very long, it introduces a lot of different problems that are not really specific and that can lead to several different huge discussions (for example: dimensionality reduction, distance between object, data visualisation, clustering, ...). I guess a paper for such a discussion should be more specific ? $\endgroup$ Dec 14, 2010 at 18:07
  • $\begingroup$ @robin Good point, but I think it is good -- I don't know how many people will come, so more chances for discussion means less chance for deadly silence ;-) $\endgroup$
    – user88
    Dec 14, 2010 at 20:11
  • $\begingroup$ There is an other problem -- the author works for US Navy and seemingly his e-mail is confidential... $\endgroup$
    – user88
    Dec 14, 2010 at 20:23
  • $\begingroup$ @mbq - unless this is the same Jeff L. Solka?? binf.gmu.edu/~jsolka $\endgroup$
    – onestop
    Dec 14, 2010 at 20:30
  • $\begingroup$ @onestop Oh, thanks. $\endgroup$
    – user88
    Dec 14, 2010 at 20:40

How about an article from Statistical Surveys, an open-access peer-reviewed journal that publishes what its title suggests?

This one in the current volume caught my eye for some reason:

  • Arlot S, Celisse A. A survey of cross-validation procedures for model selection. Statistical Surveys 2010; 4:40-79. DOI: 10.1214/09-SS054

-- but I haven't even checked the contents pages of the previous volumes so i'm open to other suggestions, especially ones with fewer than 40 pages.

  • $\begingroup$ (+1) This one is particularly theoretical (IMO), and hard to read for people with few mathematical statistics background. But SS has very good papers, as Statistical Science does. I need to check myself. $\endgroup$
    – chl
    Dec 12, 2010 at 21:17
  • $\begingroup$ Thanks for suggestion; topic looks nice, but the size is be quite discouraging. $\endgroup$
    – user88
    Dec 12, 2010 at 21:20
  • $\begingroup$ @mbq Statistical Science is a longer-established review journal but isn't open access. Sorry hadn't read past the intro of that one, and doing so now, agree it does rapidly get too theoretical-mathematical. @mbq agree about the size too! Maybe I should withdraw this specific article as a suggestion. Journal still looks a promising source of open-access review articles though. $\endgroup$
    – onestop
    Dec 12, 2010 at 22:46

You must log in to answer this question.