Ideas on A Really Fast Statistics Journal | Yihui Xie | /en/2012/03/a-really-fast-statistics-journal/

yihui 2022-12-16 21:42:17

Ideas on A Really Fast Statistics Journal | Yihui Xie

https://yihui.org/en/2012/03/a-really-fast-statistics-journal/

6 Comments

giscus-bot 2022-12-16 21:42:18

Guest *Jian-feng Mao* @ 2012-03-16 09:15:41 originally posted:

Dear Yihui, Very Excellent Ideas and applicable tech guides. I agree well with you, though I am not working in pure stat area, but in Biology. The same issues you pointed out here, occurred also for Biology and for all other academic publishing areas (I guess). We definitely need some improvements in formatting, data/code sharing, and also peer review process.
In BioMed area, f1000research (http://f1000research.com/) has made the first step. "F1000 Research will offer immediate publication; open, post-publication peer review; open revisioning of work including ongoing updates;and encourage raw data deposition and publication. In addition, F1000Research will accept a broad range of article formats and content types." I think F1000research, by its post-publication peer-review rule, would be an ideal one fit your expectation (fast publication). Do you want to express any of your ideas on post-publication peer-review? I have not found them in this blog, though I have felt your passions on that.Jian-feng,

yihui 2022-12-16 21:42:24

This is the first time that I have written something on publication, so you will not find anything else in my website. For peer review, we should do it before a paper is published (ideally by a small number of reviewers but other people also have the opportunities to review it -- just like F1000); what is equally important is comments from general readers to eliminate the possible bias of previous reviewers and show the importance of the work (readers can vote). F1000 mentioned the concern about the quality of comments, and I think this can be easily solved by public voting, e.g. with a commenting system like Disqus, you can hit "Like" on a comment, then we can order the comments by how many likes they have received. In all, we rely on "natural selection" (either selection of good reviewers or comments).

Originally posted on 2012-03-17 05:54:06

giscus-bot 2022-12-16 21:42:19

访客 *大學內民明丸* @ 2012-03-19 03:25:49 写道：

D3 is really cool.

yihui 2022-12-16 21:42:25

Yes, it will be even cooler if D3 can be embedded directly in papers :)

Originally posted on 2012-03-19 05:23:38

giscus-bot 2022-12-16 21:42:26

访客 *大學內民明丸* @ 2012-03-21 09:53:54 写道：

Yeah, I may discuss with ICSA Bulletin's editor, to use D3 in next issue.

giscus-bot 2022-12-16 21:42:20

Guest *Carl Boettiger* @ 2012-03-24 00:01:59 originally posted:

Yihui, great piece, I agree entirely with your emphasis on markdown, but a few thoughts -- one thing LaTeX actually does handle nicely is citations. Since I'm already running my document through knitr, I've thought of generating citations on-the-fly in knitr (this could be done by giving the doi to crossref api, as I mention http://www.carlboettiger.info/archives/4259), or with a bibtex maybe?

Also, web stats are useful, but so are citations. Citations are linked data, and I believe that ideally we should be making up our citations with the reason for the citation (supports, refutes, etc), using something like the CiTO ontology: http://speroni.web.cs.unibo.it/cgi-bin/lode/req.py?req=http:/purl.org/spar/cito

What do you think? I've been meaning to give this a whirl in knitr sometime, but with your talent we could probably have a more elegant solution than I'd hack together for easy and intelligent citations for markdown pubs.

yihui 2022-12-16 21:42:26

I think pandoc is good enough for dealing with citations. As I have started eating my own dog food, I will see how this actually works soon. I have another hackish package named tweakr which aims to generate bib files for R packages ( https://github.com/yihui/tweakr/blob/master/R/citation.R ), but I have not worked on it for a long while. I had this idea since the initial design of knitr: yihui/knitr#13 I hope to test it out in the near future.

I agree citations are good measures of the influence of papers, and web statistics can be good auxiliary information as well (with more details).

There are a few small steps to go before I'm satisfied with knitr with markdown, and they will be done in version 0.5.

Originally posted on 2012-03-25 06:04:49

giscus-bot 2022-12-16 21:42:27

Guest *Carl Boettiger* @ 2012-03-25 07:16:36 originally posted:

Neat, didn't realize pandoc could do non-latex citations -- will have to look into that. tweakr sounds kinda like the write.bib() from the bibtex package? As I mentioned on the knitr list, I just put together a little example of what I was thinking of, much like your issue idea, it auto-generates a citation list: https://github.com/cboettig/knitcitations/blob/master/inst/examples/citations.md). I like where your going with this markdown stuff, looking forward to seeing the latest in 0.5.

yihui 2022-12-16 21:42:28

This is much more sophisticated than I expected and is an interesting way to go. I was mainly thinking of citing R packages only.

Originally posted on 2012-03-25 19:00:06

giscus-bot 2022-12-16 21:42:29

Guest *Harlan* @ 2012-04-15 14:34:51 originally posted:

This is super-cool, but it'd be even super-cooler if the writer could give an abbreviated name for a DOI on first use (or in a header or something), so that the Markdown is readable:

citep(HalpernEtAl2006="10.1111/j.1461-0248.2005.00827.x")and thereafter...citep("HalpernEtAl2006")

giscus-bot 2022-12-16 21:42:30

Guest *Carl Boettiger* @ 2012-04-16 16:56:51 originally posted:

@openid-43133:disqus, interesting idea, I've just implemented something like this based on your suggestion. You would have to give the argument as a list the first time:

citep(c("HalpernEtAl"="10.1111/j.1461-0248.2005.00827.x") and then can use `citep("HalernEtAl") thereafter. That will keep the source mardown more readable. Recall in knitr that the output can be markdown as well, in which case it will already appear fully formatted.

giscus-bot 2022-12-16 21:42:21

Guest *dwskau* @ 2012-06-18 21:12:14 originally posted:

This is needed for every discipline, not just statistics journals.

giscus-bot 2022-12-16 21:42:22

Guest *Joshua Wiley* @ 2012-07-08 23:25:41 originally posted:

This was amazing. I agree with basically every point you make, and I would love to see this implemented for everyone (but particularly for psychology). Perhaps, better than making it restricted to one discipline, you make it like PLoSONE? Accept all papers and have tags for different fields?

yihui 2022-12-16 21:42:30

Yep, we may really do it. I have a student working on Google Summer of Code, and I'll see how far we can go in this summer. Thanks!

Originally posted on 2012-07-09 00:44:24

giscus-bot 2022-12-16 21:42:23

Guest *essay service* @ 2016-01-03 12:06:31 originally posted:

This technique that you gave will surely help a lot of people in creating their journal to become effective and will be finished in a short period of time. This will totally be applicable for many students.