Seeding the network

[mikeg...@gmail.com]

This is the thread to discuss ideas for promoting the tool, kicking of
new review networks, etc.

[mikeg...@gmail.com]

A lot of people have pointed out that this project could have
usefulness outside of academics. I think it has the most to offer in
academics, but what do you guys think about trying to seed a network
in some other area? For example, armature musicians could use it to
try to create a name for themselves independent of the big media
industries. Could tools be created to encourage this use? As long as
it started to grow, it seems likely encourage the eventual bleeding
into academia.

[UBfusion]

I would much prefer GPeerReview to remain in the academic perspective.
Everything else is already conceived and applied. Musicians have
already discovered web 2.0 and publicise their work in e.g. Facebook
or Myspace.

In fact, almost anything that has commercial interest has already been
discovered. GPeerReview is fresh, non-profit, scientifically oriented,
and I sense much needed in the knowledge society. Isn't this enough
for you?

[mikeg...@gmail.com]

Here's some action-items for which we need some discussion and
volunteers.

1- It seems that people are demanding a graphical interface. I
personally prefer a command-line interface, but I think I must give in
here. We should start a thread to discuss the features and design that
will go into it. (For example, Should it be a client app or run in a
browser? Should there be one tool for signing and a separate tool for
graph analysis? How do you envision the interface? Should it support
plug-ins for multiple graph-analysis algorithms? Who wants to begin
writing it? etc.) After the design starts to gel, we can begin
development. Whoever writes the code, of course, will have the last
word on design issues, unless someone else wants to fork it. I am an
experienced coder, but I'd rather hand off to someone else if anyone
is interested, so I can focus on coordinating efforts.

2- I've heard a lot of theoretical discussion, but I haven't heard
much discussion about the tool itself. Can anyone who has actually
comment about its current state of usability and needed features?

3- I have become increasingly persuaded that the only way for this
tool to succeed is if we can convince a few conferences or journals to
begin using it. Only then will people take it seriously. I think a
major part of our promotional efforts should center around finding
venues that might adopt it, and adapting the tool (but not the
fundamental design principles of course) to their liking. So, we need
some ideas for how to organize such an effort.

4- It has been proposed that we should seek for grants to assist this
project. Nathan will do a preliminary glance for relevant grants. It
would help if some more people helped look. Any volunteers?

5- It has been suggested that we should contact some professors who
might encourage graduate students to develop analysis algorithms for a
thesis project. Does anyone have contacts in relevant fields?
Cryptography? Data-mining? Graph theory? Information retrieval?

6- Would anyone volunteer to look into integrating with DOI, or arXiv,
or related systems?

Thankyou everyone for all the interest in this project.

On Feb 5, 8:06 am, mikegash...@gmail.com wrote:

[Qubyte]

I imagine that analysis tool will be separate. One thing that will
make crawling the review network easier is with RDF, perhaps in RSS
feeds. There could be a web based tool (I'm thinking based on app
engine) that keeps a seed list of RSS feeds and does a regular crawl.
It should of course also check EO's for new research that isn't
connected yet, although probably the best thing to encourage is for a
new researcher to submit some reviews when they perform a literature
review. That way they're already connected and the crawler tool can
find them. Part of the tool (for some gloss) can be a visual
representation of you in the graph. That's simple to do using networkx
or boost. Note that both don't work with app engine but networkx can
probably be tweaked to by removing a dependency. The tool is going to
have some problems, but those probably won't surface until it's being
tested. Otherwise I guess that most of the work is in developing an
algorithm to weight researchers. There's some interesting stuff in
this paper:

http://arxiv.org/abs/cond-mat/0106096

I'm a quantum information scientist. It's a pretty new field, with a
lot of young blood. People in it seem to like adopting new tools, so I
think this community is primed for picking this up. I imagine there
are some other young fields that are the same. You'll only need to get
them enthusiastic and related fields join in pretty quickly. What we
will need though is a really solid specification for doing reviews,
and some servers working for us. I'm looking around at the moment for
some funding to set up a quantum information EO as a prototype. I'm
still in the field of course so I would promote the idea.

One question is the exact role of the EO. Frankly I'm not sure I trust
people not to lose their key, or keep their personal website up. If
one goes down, then that could cause all sorts of problems. It may be
worth giving everyone who joins an EO the option of a free page just
for their publications etc. On the other hand maybe a completely
different server is a better idea. A data organisation (DO) which
gives you a little page for free which you can edit and link to your
institution page would be cool, and then it can also store a copy of
your paper for checking purposes.That doesn't remove the option for
self publication, but it does give you one that is likely to be more
robust and won't change when you move to another uni.

Any thoughts on this?

On Feb 6, 5:54 pm, mikegash...@gmail.com wrote:
> Here's some action-items for which we need some discussion and
> volunteers.
>
> 1- It seems that people are demanding a graphical interface. I
> personally prefer a command-line interface, but I think I must give in
> here. We should start a thread to discuss the features and design that
> will go into it. (For example, Should it be a client app or run in a

> browser? Should there be one tool for signing and a separate tool forgraphanalysis? How do you envision the interface? Should it support
> plug-ins for multiplegraph-analysis algorithms? Who wants to begin

> writing it? etc.) After the design starts to gel, we can begin
> development. Whoever writes the code, of course, will have the last
> word on design issues, unless someone else wants to fork it. I am an
> experienced coder, but I'd rather hand off to someone else if anyone
> is interested, so I can focus on coordinating efforts.
>
> 2- I've heard a lot of theoretical discussion, but I haven't heard
> much discussion about the tool itself. Can anyone who has actually
> comment about its current state of usability and needed features?
>
> 3- I have become increasingly persuaded that the only way for this
> tool to succeed is if we can convince a few conferences or journals to
> begin using it. Only then will people take it seriously. I think a
> major part of our promotional efforts should center around finding
> venues that might adopt it, and adapting the tool (but not the
> fundamental design principles of course) to their liking. So, we need
> some ideas for how to organize such an effort.
>
> 4- It has been proposed that we should seek for grants to assist this
> project. Nathan will do a preliminary glance for relevant grants. It
> would help if some more people helped look. Any volunteers?
>
> 5- It has been suggested that we should contact some professors who
> might encourage graduate students to develop analysis algorithms for a
> thesis project. Does anyone have contacts in relevant fields?

> Cryptography? Data-mining?Graphtheory? Information retrieval?

[Qubyte]

As an example of RDF which should be very familiar:

http://en.wikipedia.org/wiki/FOAF_(software)

On Mar 12, 10:42 am, Qubyte <mark.s.ever...@gmail.com> wrote:
> I imagine that analysis tool will be separate. One thing that will
> make crawling the review network easier is with RDF, perhaps in RSS
> feeds. There could be a web based tool (I'm thinking based on app
> engine) that keeps a seed list of RSS feeds and does a regular crawl.
> It should of course also check EO's for new research that isn't
> connected yet, although probably the best thing to encourage is for a
> new researcher to submit some reviews when they perform a literature
> review. That way they're already connected and the crawler tool can
> find them. Part of the tool (for some gloss) can be a visual

> representation of you in thegraph. That's simple to do using networkx

[Andrew Taylor]

A couple of days ago I chanced upon http://www.mendeley.com/

Mendeley is a reference manager, but with web and social features,
including an author profile with a list of papers (optionally
including downloadable papers). It seems to provide a ready-made
network and interface that GPeerReview could easily exploit.

It would have to be cooperative, though. Their business model
(whatever it is) may or may not not be suited to GPeerReview, but they
are keen on open-access so their philosophy might be very compatible.

[Qubyte]

Has anyone contacted arXiv about this? I trust arXiv not to be evil,
and it enjoys deep integration with the research community. It has
some architectural issues, such as rebuilding documents now and then,
but it gpeerreview takes a hash of the LaTeX source code rather than
the compiled document this shouldn't be a problem. Also versioning
could be an issue here.

If nobody has done this then I'll give it a shot.

On Mar 12, 10:51 am, Andrew Taylor <taylor.and...@gmail.com> wrote:
> A couple of days ago I chanced uponhttp://www.mendeley.com/

[Mike Gashler]

Here's a reply I got in Aug '08 from arXiv.org. We've evolved a lot since then, though, so it would be nice to hear from them again.

Dear Mike,

Thank you for message.  We may consider something like this in the future.  It is certainly an interesting notion for supporting overlay journals.

--
arXiv admin

Mike Gashler wrote:
> Dear arXiv,
>     I am the author of a tool that facilitates digitally-signing peer reviews. (http://code.google.com/p/gpeerreview/) In my opinion, this tool compliments the excellent service that arXiv already provides. It would be really great if you guys were to think about incorporating something like this into your service.
>
> thanks,
> -Mike
>

I recently submitted an abstract about GPeerReview to ISPR'09 (http://www.iiis2009.org/wmsci/website/default.asp?vc=27). I'm hoping that if we can publish something in a refereed conference about GPeerReview, that might help to get the attention of serious players like arXiv.org, DOI, etc.

Regarding separate analysis tools, I agree that it makes sense to have separate analysis tools. We'd like to encourage people to use different analysis techniques so there is not a single formula for people to try to game.

Regarding RDF, I'm not 100% clear where you are proposing to use it. I think you're suggesting to embed RDF within the signed endorsements for the purpose of referencing the reviewer's hompage and c.v. That could work. I don't yet see much advantage over just a plain-old simple hyperlink. We should certainly try to incorporate standards as much as possible as long as they don't limit or significantly complicate what we want to do.

Regarding people being sloppy with their private keys: This is somewhat of a problem with all cryptography in general. If the keys are in untrustworthy hands, it doesn't matter how strong the lock is. I think the primary mitigating factor with our system is that the cryptography isn't really even strictly necessary for it to work. The cryptography is really just a time-saving mechanism. It enables the automation of processes that would otherwise require a phone call or the sending of an email. The current (non-automated) system works (to some extent) without cryptography. A dishonest person can list all sorts of phony publications on his c.v., and get away with it until people start talking to each other. I think the same checks will ensure that even the sloppy handling of private keys won't ultimately do much damage. If we were transmitting money to/from an untrusted person, this weakness would be a really big deal. In this case, however, someone is trying to establish credibility, probably with the intent of getting a job or seeking tenure. If you try to pull that with phony endorsements that you made with a stolen private key, you'll almost certainly be caught within a few years anyway. You can't exactly get tenure and then flee the country with it.

[Qubyte]

Yeah, there isn't really a solution to the loss of keys and I can't
see a way around that. The point I guess I was trying to make with RDF
is that GPeerReview will construct networks that look a lot like FOAF.
Tim Berners-Lee has been going on about the GGG, and this kind of
network seems to be closely related so there may be some tools about
that we could make useful for our purposes.

Getting arXiv on board would be a boon. If you told people that GPR
was already on arXiv for them to use then you'd see an explosion of
use. Mike, do you fancy writing a paper to put on the arXiv? This
would be a fantastic way of grabbing attention. I'll happily
contribute to this if you like. I'm a dab hand with LaTeX.

[Mike Gashler]

Putting a paper on arXiv about it is a great idea. I'm afraid it might take me a long time to get this done since I'm pretty swamped with my Ph.D. dissertation on another topic, and when I find time for GPeerReview, I'd like to use it to develop the tools. Would you or anyone else like to be the first author?

[Qubyte]

Same here! I'm guessing I'll be neck deep in thesis for the next month
or so, but after that I'm game. I may start drafting something in the
next fit of procrastination.

[Matthews]

Dear gpeerreview,

Could this system be tested on a series of Wikipedia entries, as a test, as public proof of concept, as a way of gainign publica recognition, and also to eventually provide a comprehensive rating system for Wikipedia entries?

This would be much easier than persuading conference organisers to get permissions from participants to have their work subjected to unknown, unseen review system.

Cheers, Peter

***

--
http://cooperative.ning.com

An international, online meeting place for research writers, editors, translators, and publishers

[Qubyte]

Wikipedia would be a really cool proving ground, but it lacks author
culpability since many authors can contribute to a page. In addition
the article may evolve considerably between reviews.

> --http://cooperative.ning.com