refdb-users

[Refdb-users] Re: interface mockup

[Bruce D'A. <bd...@fa...>]

On Jan 30, 2004, at 9:35 PM, Matt Price wrote:

> On Fri, Jan 30, 2004 at 12:20:34PM -0500, Bruce D'Arcus wrote:
>> I don't think this made it to the list.  Any feedback?  I can mod the
>> xslt once I have a better sense of what we want.
> Though I can see that other folks might not want this, I'd like to
> make a plea for a hot-linked title in a short-format list.

So what would happen if you click on the link?

> This is mostly because once you have an <a href> tag, it's possible to 
> write
> fully-functional mycroft search engines for mozilla (see
> http://mycroft.mozdev.org)...

Interesting; I'd not heard about that.  I wonder if the 
browser-agnostic bookmarklet might be better though?  E.g.:

http://webvoy.uwindsor.ca:8080/cocoon/mount/chrysalis/welcome

> So, missing from here is still the code for _adding_ references and
> notes, correct?  How do you think this should be done?

A series of template defined fields that map natural language names to 
underlying metadata model/format.

> Notes should be relatively simple -- the only hard part is getting the
> information into the xml format that refdb uses.  If poss I'd like to
> see this done in xslt again, so that the hard part is abstracted from
> the scripting language and can be recycled in perl/python/whatever.
> Bruce, you're the xslt person, how does that shound to you?

I don't do scripting, so maybe Rob has a better sense of the connection 
between the XML and a form interface.

> 1) this is important -- choosing a reference format.  Do you think risx
> is hte way to go again, or would something simple like bibtex be more
> appropriate?

I despise bibtex myself.  I think the input UI should be designed for 
flexibility; it should work elegantly with risx, but be capable of 
being moved to handling mods data later.  The trick is to define the 
key user input fields in a generic way, and then allow configuration.  
So, the generic fields might be:

creator/editor
title/subtitle
year
publisher
origin place (could be for unpublished sources too)
location
keywords
etc.

> 2) I'm a bit worried about parsing author fields

There's no doubt this is the trickiest part of any form-based UI for 
bib data.  I'm not sure how it ought to work.  Anyone have any ideas?  
Rob?

Bruce

[Refdb-users] Re: interface mockup

[Bruce D'A. <bd...@fa...>]

I like this addition.  I also like the paste-in-the-field thing in the 
current refdb web interface, which is good when doing online searches 
and such.

On Jan 31, 2004, at 12:24 AM, Rob Casson wrote:

> i added an ability to upload ris data via the web the other day..didn't
> send it to the general list, just the refdb-dev...it was more of an
> aside appended to a different concern...anyway, feel free to try it
> out:
>
>     http://junk.lib.muohio.edu/refdb/upload.php

Re: [Refdb-users] Re: interface mockup

[Matt P. <mat...@ut...>]

On Fri, Jan 30, 2004 at 10:16:30PM -0500, Bruce D'Arcus wrote:
> On Jan 30, 2004, at 9:35 PM, Matt Price wrote:
> 
> >On Fri, Jan 30, 2004 at 12:20:34PM -0500, Bruce D'Arcus wrote:
> >>I don't think this made it to the list.  Any feedback?  I can mod the
> >>xslt once I have a better sense of what we want.
> >Though I can see that other folks might not want this, I'd like to
> >make a plea for a hot-linked title in a short-format list.
> 
> So what would happen if you click on the link?

you get a page with a full reference, including notes and abstract &
whatever else...

> 
> >This is mostly because once you have an <a href> tag, it's possible to 
> >write
> >fully-functional mycroft search engines for mozilla (see
> >http://mycroft.mozdev.org)...
> 
> Interesting; I'd not heard about that.  I wonder if the 
> browser-agnostic bookmarklet might be better though?  E.g.:
> 
> http://webvoy.uwindsor.ca:8080/cocoon/mount/chrysalis/welcome
> 
hmm.  I see the advantages of agnosticism here (and I'm on console
right now, so haven't checked the link) but mycroft has the advantage
of making it possible to combine searches from multiple engines; for
bibliographic information this is not yet especially useful, since
complex queries are rendered with different syntax on different sites;
but eventually I think a more flexible interface will be implemented.
This can be a substantial advantage when data is distributed over a
variety of sites.  

> >So, missing from here is still the code for _adding_ references and
> >notes, correct?  How do you think this should be done?
> 
> A series of template defined fields that map natural language names to 
> underlying metadata model/format.

...  ok, as marcus said recently, how does that translate into
language that a plain old PhD like myself can understand?  though I
actually don't intend any snide overtones that might be inadvertently
comunicated.  when you say 'map natural language', I take it you mean
"take data entered in fields whose names are readily comprehensible
to users, and pass the data on to the database in fields whose names
the db understands." but what's the importance of "template-defined"?

> 
> >Notes should be relatively simple -- the only hard part is getting the
> >information into the xml format that refdb uses.  If poss I'd like to
> >see this done in xslt again, so that the hard part is abstracted from
> >the scripting language and can be recycled in perl/python/whatever.
> >Bruce, you're the xslt person, how does that shound to you?
> 
> I don't do scripting, so maybe Rob has a better sense of the connection 
> between the XML and a form interface.
> 
Rob, tell me what you think -- I think we're interested in similar
goals here, maybe we can divide up the work a bit.  

> >1) this is important -- choosing a reference format.  Do you think risx
> >is hte way to go again, or would something simple like bibtex be more
> >appropriate?
> 
> I despise bibtex myself.  I think the input UI should be designed for 
> flexibility; it should work elegantly with risx, but be capable of 
> being moved to handling mods data later.  The trick is to define the 
> key user input fields in a generic way, and then allow configuration.  
> So, the generic fields might be:
> 

Though I like the way that sounds, it sounds pretty ambitious.  For
now it might be much less labour-intensive to write some php that
generates bibtex or other simple data structures; though again, my
pref would be to have the interesting part abstracted as much as
possible from the scripting language.  

> creator/editor
> title/subtitle
> year
> publisher
> origin place (could be for unpublished sources too)
> location
> keywords
> etc.
> 
> >2) I'm a bit worried about parsing author fields
> 
> There's no doubt this is the trickiest part of any form-based UI for 
> bib data.  I'm not sure how it ought to work.  Anyone have any ideas?  
> Rob?
> 
> Bruce
> 

Re: [Refdb-users] Re: interface mockup

[Bruce D'A. <bd...@fa...>]

On Jan 30, 2004, at 11:26 PM, Matt Price wrote:

>> So what would happen if you click on the link?
>
> you get a page with a full reference, including notes and abstract &
> whatever else...

How would that then relate to the mark-and-view/export interface, and 
note pop-up buttons?  Do me a favor and compare what you're thinking 
about to this example:

http://130.83.186.203/exist/xquery/biblio.xml

>>> So, missing from here is still the code for _adding_ references and
>>> notes, correct?  How do you think this should be done?
>>
>> A series of template defined fields that map natural language names to
>> underlying metadata model/format.
>
> ...  ok, as marcus said recently, how does that translate into
> language that a plain old PhD like myself can understand?  though I
> actually don't intend any snide overtones that might be inadvertently
> comunicated.

Well ... if you wanted to avoid being snide you could have just said "I 
don't understand; please explain." ;-)

>  when you say 'map natural language', I take it you mean
> "take data entered in fields whose names are readily comprehensible
> to users, and pass the data on to the database in fields whose names
> the db understands." but what's the importance of "template-defined"?

OK, you have a pop-up list with reference types of, for example:

book
edited book
journal article
legal case
film
etc.

You then choose "create new record" and a form specific to that type is 
then displayed.  The template -- maybe an XML configuration file -- 
simply says (using ris stuff):

JOUR = "journal article"
T1 = "title"
t2 = "journal title"

..and so forth.  It's just an abstraction that makes it a) easier for 
the user and b) easier to later modify/add reference types.

>>> 1) this is important -- choosing a reference format.  Do you think 
>>> risx
>>> is hte way to go again, or would something simple like bibtex be more
>>> appropriate?
>>
>> I despise bibtex myself.  I think the input UI should be designed for
>> flexibility; it should work elegantly with risx, but be capable of
>> being moved to handling mods data later.  The trick is to define the
>> key user input fields in a generic way, and then allow configuration.
>> So, the generic fields might be:
>
> Though I like the way that sounds, it sounds pretty ambitious.  For
> now it might be much less labour-intensive to write some php that
> generates bibtex or other simple data structures; though again, my
> pref would be to have the interesting part abstracted as much as
> possible from the scripting language.

Well, if you want to avoid XML, use RIS (it's richer, and 
unicode-possible).  But Rob, doesn't the XML support in PHP allow you 
to basically say "take date in field X and create XML code Y"?  If yes, 
that'd be preferable.

Bruce

Re: [Refdb-users] Re: interface mockup

[Bruce D'A. <bd...@fa...>]

On Jan 31, 2004, at 9:22 AM, Bruce D'Arcus wrote:

>>>> Though I can see that other folks might not want this, I'd like to
>>>> make a plea for a hot-linked title in a short-format list.
>>> So what would happen if you click on the link?
>>
>> you get a page with a full reference, including notes and abstract &
>> whatever else...
>
> How would that then relate to the mark-and-view/export interface, and 
> note pop-up buttons?

Here's an idea:

Click on the title, and a window a la the Moveable Type comment window 
(e.g. small and plain) pops up with the full display (metadata plus 
notes).   We'd keep the mark facility to handle multiple records.

Another idea following from this (which I'm unsure on) is to also build 
on blog commenting by adding a note field at the bottom of the detailed 
display window (???).  If that was the case, the note buttons on the 
main result display would go away I think.

Bruce

Re: [Refdb-users] Re: interface mockup

[Bruce D'A. <bd...@fa...>]

On Feb 3, 2004, at 12:08 AM, Matt Price wrote:

> On Sat, Jan 31, 2004 at 09:22:32AM -0500, Bruce D'Arcus wrote:
>> On Jan 30, 2004, at 11:26 PM, Matt Price wrote:
>>
>>>> So what would happen if you click on the link?
>>>
>>> you get a page with a full reference, including notes and abstract &
>>> whatever else...
>>
>> How would that then relate to the mark-and-view/export interface, and
>> note pop-up buttons?  Do me a favor and compare what you're thinking
>> about to this example:
>>
>> http://130.83.186.203/exist/xquery/biblio.xm
>
> xquery looks interesting -- I will check it out Real Soon Now.

The thing to look at is just the interface.

>>> when you say 'map natural language', I take it you mean
>>> "take data entered in fields whose names are readily comprehensible
>>> to users, and pass the data on to the database in fields whose names
>>> the db understands." but what's the importance of "template-defined"?
>>
>> OK, you have a pop-up list with reference types of, for example:
>>
>> book
>> edited book
>> journal article
>> legal case
>> film
>> etc.
>>
>> You then choose "create new record" and a form specific to that type  
>> is
>> then displayed.  The template -- maybe an XML configuration file --
>> simply says (using ris stuff):
>>
>> JOUR = "journal article"
>> T1 = "title"
>> t2 = "journal title"
>>
> I think I understand this.  That should be relatively straightforward
> using php...

Take a look here for something like what I'm talking about.

http://216.205.78.164:8001/bw_login2srch? 
edit_loginname=guest&edit_password=guest

For some reason I can't create new references, now, so can't see the  
form.  Otherwise, the interface is pretty nice.

Another thing to look at that I'm quite fond of, which uses Cocoon and  
the XML DB eXist, and whose author is working on updating it to support  
MODS, is the MetaWizard for the SozioNet project out of Germany.   
Screenshots on my blog.

http://netapps.muohio.edu/movabletype/archives/darcusb/darcusb/ 
000129.html

Bruce

Re: [Refdb-users] Re: interface mockup

[Markus H. <mar...@mh...>]

Matt Price writes:
 > Though I like the way that sounds, it sounds pretty ambitious.  For
 > now it might be much less labour-intensive to write some php that
 > generates bibtex or other simple data structures; though again, my
 > pref would be to have the interesting part abstracted as much as
 > possible from the scripting language.  
 > 

Expressing data in XML is less labour-intensive than you may
think. RefDB (it is written in C, but the principles are the same in
any scripting language) for example uses four helper functions to
create risx XML output:

print_element_x: generates a complete element including contents and
                 attributes. It uses print_elstart_x and print_elend_x
                 internally.

print_elstart_x: generates the start tag including the attributes. It
                 uses print_attribute_x internally.

print_elend_x:   generates the end tag

print_attribute_x: prints out an attribute name/value pair

All you need to figure out is how to call these functions with your
data in order to create valid output. However, the complexity
of risx and xnote is limited. It's doable.

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

[Refdb-users] Re: interface mockup

[Markus H. <mar...@mh...>]

 > >> I don't think this made it to the list.  Any feedback?  I can mod the

I'm sorry but the SF mailing lists limit the size of incoming mail. If
a post exceeds this limit, it will be in the loop until I, as the list
administrator, approve it. Hence the delay of up to 24h. If you have
the means to, please put larger files on your web space and post the
link instead.

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de