Journal

[Tim Jurka]

Perhaps I'm being optimistic, but isn't Political Analysis ( http://pan.oxfordjournals.org/ ) the appropriate place for such an article?

Tim

[Jon Moody]

Hey all,

I've been quiet for quite a while on all of this as I've been sidetracked working on my dissertation; however, it might make sense to do a pair of articles for the new version of TextTools.  One is the introductory article that showcases the software and the latter is an application strategy article, for which I already did a majority of the leg-work using the original version of TextTools.  Would it not make sense to submit the articles in tandem?  If so, I realistically could have the second one finished or at least ready for the rest of the co-authors to review sometime in July/August.  Thoughts?

- Jon

[Amber Boydstun]

Hey Jon,

Good thought.  I think we're all traveling at the moment, but Emiliano and I will touch base with Loren and Wouter and Tim in the next few weeks to get their thoughts and then we can come up with a publishing game plan.

Cheers,
Amber

~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Amber E. Boydstun
Assistant Professor
Department of Political Science
University of California, Davis
One Shields Ave
Davis, CA 95616

http://psfaculty.ucdavis.edu/boydstun/
~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[Wouter van Atteveldt]

Dear all,

I probably won't be around for the conversation as I am busy cycling from (hopefully) tomorrow.

I think that the more polsci-oriented journals will indeed need more of an application in addition to presenting the texttools package. I think PolAna will also need a major methodological contribution, which RTextTools by itself isn't (as the techniques themselves are not new).

I am working on two Machine Learning pieces at the moment, one resubmission of my CAP paper last year (on ensemble versus probability) and a writeup of my Text as Data paper of this march (on multilingual ML). so I would like to be actively involved if the paper touches these topics.

If the application is in another domain, or if we decide to do just a description, I am happy to be on the back burner and help with the writeup et cetera from August.

Personally, I would think we should not attempt to turn this into a very high-ranked publication, as this would involve significant extra work, in which case I think it is best to submit it first as a plain description, and then work on the major publication. So, I think there are two feasible options for the first publication:

1) Polish the API, write up the description, and submit asap to J Stat Softw

2) Write a piece with a simple application aimed at the PolSci audience, and submit to a methodologically focussed PolSci journal

In both cases, as the work is mainly theirs and they are probably most in need of a first-authored methodological publication, I think Tim and Loren should take the initiative and one of them should be first author. In any case, all of us should be on the first publication.

Enjoy Sicily for a bit, and I'll try to join the publication discussion on the mailing list...

-- Wouter

[Tim Jurka]

Since there isn't much original statistical work in RTextTools, Loren and I think it'd be more appropriate to submit the initial paper to RJournal, and then the subsequent paper in a methodologically focused journal. Thoughts?

Tim

[Loren Collingwood]

I'm thinking a quick RJournal would be nice, especially since I'm presenting RTextTools at the UseR! conference in England in August. So Tim and I will work on something and pass around (hopefully) shortly.

-Loren

Loren Collingwood
Ph.D. Candidate
Department of Political Science
University of Washington
http://staff.washington.edu/lorenc2
lor...@uw.edu

[Jon Moody]

That makes sense to me.  I just know that the paper I've been working on for forever and a day that used the original TextTools was designed to introduce it to social scientists (namely political scientists) as a viable tool that they can use.  There was an issue of PolAnalysis that highlighted a bunch of the automated text techniques (i.e. Shrodt and Monroe methods), but was still largely inaccessible for the average scholar.  It seemed more like the authors showing off than presenting a tangible method for others to use.  So my approach was to simplify it a bit, refer people to documentation and example data to familiarize themselves with the software, and then offer some strategies for how to best apply the method to their data, while taking some of the dangers of the method into account.

- Jon