refdb-users — List for RefDB users

[Refdb-users] Endnote to RIS to RefDB troubles

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

Hi,

Marc Baaden writes:
 > An example ris file that reproduces those can be found at
 > http://baaden.nerim.net/pub/err.ris
 > 
 > Being rather a newby, I wonder what is going wrong and how to
 > correct it ?
 > Does the Endnote RIS export filter need some updating, or
 > should I write a perlscript to tweak some probably malformed
 > entries ?
 > Obviously with 4000 references, you don't want to correct the
 > several hundred errors by hand :)) (To be precise, 3400 dataset(s)
 > added, 352 failed)
 > 

I'll have a look at the ris file. I've experienced a few problems with
the 0.9.2a release myself, all of them were related to bugs in the
auto-creation of citation keys (the bugs are described on the refdb
project page at sourceforge). These bugs have been addressed in the
CVS version. I'll find out whether upgrading would help in your case.

 > 
 > == Keeping IDs ?? ==
 > 
 > As I have documents which reference the bibliographic entries
 > by the actual ID numbers from Endnote, I would like to keep
 > that numbering. I have tried with the -k option like in
 > 
 > refdbc -d bibthese -C "addref -k Biblio_THESE.ris"
 > 
 > But get the following errors continuously:
 > 
 > ID 19 ignored, I have assigned a new one
 > try to add set as reference 293
 > 
 > What am I doing wrong ?
 > 

There is currently no way to convince the SQL database server to reuse
existing ID values (they have to be unique, so it is easier to let the
server create them). The -k switch simply puts the previous value into
the U5 field so it's not lost. You could retrieve references by their
previous ID with something like :U5:=1234. 

 > 
 > == Performance issues ==
 > 
 > The database I am converting is about 4000 references, and
 > it took more than an hour to add those references to refdb.
 > As I am currently experimenting, I might have to do this a
 > number of times until the result is acceptable. Is there a
 > way to speed this up ?
 > 

Could you provide some details about your setup? Which database
server, what computer, do you use RefDB across a network etc? I don't
have any sort of "benchmark" values available, but an hour sounds like
a lot to me. How long would EndNote take to swallow the data?

regards,
Markus

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

Re: [Refdb-users] Endnote to RIS to RefDB troubles

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

On Saturday, April 26, 2003, at 03:55  PM, Marc Baaden wrote:

> I have experienced some problems/difficulties by trying to
> convert some previous Endnote databases to refdb.

I've found the same.  Perhaps the two of us and Markus can figure out a 
solution?

> Being rather a newby, I wonder what is going wrong and how to
> correct it ?
> Does the Endnote RIS export filter need some updating, or
> should I write a perlscript to tweak some probably malformed
> entries ?

Both.  The export filter has problems, and I found I needed to change 
it significantly.  I will send you the version I have (which is on my 
other machine) later off-list.

However, Endnote itself has some bugs in its export system that cannot 
be corrected by fixing the filter alone.  The most notable problems are 
with keywords, where only the first is tagged, so you get something 
like:

KW  - first keyword
second keyword

The other significant problem is that dates are rendered differently in 
Endnote/Refer than in Reference Manager/RIS.  I'm not sure what to do 
about this, but some feedback from Markus would help.

This is where I think Markus (maybe with your help; I don't know how to 
write Perl scripts) ought to add some code to correct these problems on 
import.  There's a further issue for Mac users like me: encodings and 
line endings.

Bruce

Re: [Refdb-users] Importing RisX into RefDB

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

Bruce D'Arcus writes:
 > Are the only relevant tools here editors?  Or are you also thinking of 
 > XML parsers?  Are there no good free parsers that handle schemas?
 > 

All we need is a parser/XSLT combo that supports W3C schema. At least
Xerces/Xalan does the trick, I didn't check the other free
parsers/processors. The problem are indeed the editors. It would be
pointless to offer a free bibliography software and require people to
buy XMLSpy in order to enter data.

 > One idea I have is that if you do decide on an additional entry DTD for 
 > the current model, it could provide a sort of bridge between risX and 
 > MODS.  So risX is the low-level straight ris mapping, "risX plus" (or 
 > whatever you want to call it; BibX??) is more explicitly aimed at data 
 > entry, and MODS is the future -- more rational and ambitious -- data 
 > model.

I was thinking along the same lines. What I'd need is a DTD
representation of the MODS schema. This could be stripped down to a
MODS-lite by removing everything that risx couldn't reasonably
include. A later transition to MODS as the main data format would be
fairly painless then. A brief google search didn't unearth any free
Schema->DTD tools, though. Is there anyone out there with access to a
commercial tool that does the trick (XMLSpy can do this afaik)?

regards,
Markus

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

[Refdb-users] Endnote to RIS to RefDB troubles

[Marc B. <ba...@sm...>]

Hi,


I have experienced some problems/difficulties by trying to
convert some previous Endnote databases to refdb.


== Endnote exported RIS problems ==

When trying to import the endnote RIS export, I get the following
types of errors:

insert AU failed
insert JX failed
insert KW failed
query AU failed
query KW failed
update AB failed
update BT failed
update CP failed
update CY failed
update PB failed
update TI failed
update VL failed

An example ris file that reproduces those can be found at
http://baaden.nerim.net/pub/err.ris

Being rather a newby, I wonder what is going wrong and how to
correct it ?
Does the Endnote RIS export filter need some updating, or
should I write a perlscript to tweak some probably malformed
entries ?
Obviously with 4000 references, you don't want to correct the
several hundred errors by hand :)) (To be precise, 3400 dataset(s)
added, 352 failed)


== Keeping IDs ?? ==

As I have documents which reference the bibliographic entries
by the actual ID numbers from Endnote, I would like to keep
that numbering. I have tried with the -k option like in

refdbc -d bibthese -C "addref -k Biblio_THESE.ris"

But get the following errors continuously:

ID 19 ignored, I have assigned a new one
try to add set as reference 293

What am I doing wrong ?


== Performance issues ==

The database I am converting is about 4000 references, and
it took more than an hour to add those references to refdb.
As I am currently experimenting, I might have to do this a
number of times until the result is acceptable. Is there a
way to speed this up ?


Thanks in advance for any hints,
Cheers,

Marc


  Marc Baaden

Re: [Refdb-users] Importing RisX into RefDB

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

On Saturday, April 26, 2003, at 10:17  AM, Markus Hoenicka wrote:

> I currently hesitate to switch to MODS as it is implemented as a 
> Schema, not a DTD. Not many tools support this yet, and I can only 
> hope and pray that my favourite tool (Emacs and PSGML) will do so 
> eventually. But I agree that we should sit down and analyze how common 
> MODS data could be mapped to RIS or risx.

Are the only relevant tools here editors?  Or are you also thinking of 
XML parsers?  Are there no good free parsers that handle schemas?

One idea I have is that if you do decide on an additional entry DTD for 
the current model, it could provide a sort of bridge between risX and 
MODS.  So risX is the low-level straight ris mapping, "risX plus" (or 
whatever you want to call it; BibX??) is more explicitly aimed at data 
entry, and MODS is the future -- more rational and ambitious -- data 
model.

MODS, for anyone interested, is part of a larger comprehensive XML 
strategy at the Library of Congress.  Their key format to move MARC 
records to XML is MARCXML, and they have a Java conversion tool that 
does just this without loss.  From MARCXML, they have XSLT files that 
convert back-and-forth between MARCXML on the one hand, and Dublin Core 
and MODS (and a couple of others) on the other.

http://www.loc.gov/standards/marcxml//marcxml-architecture.html
http://www.loc.gov/standards/marcxml//

MODS has the same basic data structure as MARC (which strikes me as 
better than the ris model), but:

a) carries only a subset of the data
b) is explicitly designed for the digital and internet world
c) is now having additional structures added for bibliographic 
citations (long story, but the current version of MODS isn't well 
suited to representing something like a journal article)

It also seems that the LoC is positioning MODS as a key data format for 
the future.  For example, they are also involved in an effort to define 
a next generation version of z39.50, and are planning to set up a 
server that distributes MODS records.

	http://www.loc.gov/z3950/agency/zing/srw/background.html

 From this site, the LoC explains:

"Longer-range plan is to build an SRW/Z39.50 gateway, to allow access 
to our Z39.50 server from an SRW client. MARC records would be 
converted to MODS."

> Family duties may prevent this right now, but I hope I'll be able to 
> carve out some time next week.

Sounds good.  By that time, the LoC will hopefully release their new 
user guide to MODS, which might be helpful.

Bruce

Re: [Refdb-users] Importing RisX into RefDB

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

Hi,

Bruce D'Arcus writes:
 > Right.  This is what I had in mind; the DTD ought to help users know 
 > what kinds of data goes where, and in so doing create valid records.  
 > However you decide to do it (one DTD vs. two), any technical details 
 > (XSLT on import, for example) ought to be completely hidden from the 
 > user.  I think you've got the right idea though!
 > 

The technical stuff can easily be hidden from the user by providing a
simple input filter like there is for MARC or Pubmed data. This filter
would run the appropriate XSLT transformations.

 > If you do the two DTD approach, this does raise the question of MODS 
 > again (e.g. -- the longer term).  It might be a good idea to look 
 > closely at that before working on any additional DTD (and after 
 > solidifying risX proper).  I'm not sure what specific data structures 
 > ought to be added to RefDB to enable some of the features I'd like to 
 > see in a reference manager app, but my sense is that MODS and 
 > Kaspaliste (which uses PostgreSQL, and has tables for notes and links) 
 > are where to look.
 > 

I currently hesitate to switch to MODS as it is implemented as a
Schema, not a DTD. Not many tools support this yet, and I can only
hope and pray that my favourite tool (Emacs and PSGML) will do so
eventually. But I agree that we should sit down and analyze how common
MODS data could be mapped to RIS or risx. Family duties may prevent
this right now, but I hope I'll be able to carve out some time next
week.

regards,
Markus

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

Re: [Refdb-users] Importing RisX into RefDB

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

On Friday, April 25, 2003, at 06:28  PM, Markus Hoenicka wrote:

> The attributes would not be stored in the database. It would rather
> involve fixed translations on the ways in and out. E.g. if you send a
> PCOMM with a <name role="recipient">, then refdbd would know it has to
> map this name to A2. On the way out, the combination of "A2" and the
> reference type would have to result in the same name element that you
> put in.

and...

> An alternative approach would define several data models that catch
> all available reference types. In that case, you would create a
> <pcomm> element which would allow only <author> and <recipient>
> elements, but no <editor>.

Right.  This is what I had in mind; the DTD ought to help users know 
what kinds of data goes where, and in so doing create valid records.  
However you decide to do it (one DTD vs. two), any technical details 
(XSLT on import, for example) ought to be completely hidden from the 
user.  I think you've got the right idea though!

If you do the two DTD approach, this does raise the question of MODS 
again (e.g. -- the longer term).  It might be a good idea to look 
closely at that before working on any additional DTD (and after 
solidifying risX proper).  I'm not sure what specific data structures 
ought to be added to RefDB to enable some of the features I'd like to 
see in a reference manager app, but my sense is that MODS and 
Kaspaliste (which uses PostgreSQL, and has tables for notes and links) 
are where to look.

Bruce

Re: [Refdb-users] Importing RisX into RefDB

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

Hi,

Bruce D'Arcus writes:
 > The RIS spec itself.  The spec is quite clear about what each specific 
 > field is to be used for.  A book editor is A2.  Likewise, a recipient 
 > is -- or rather should be -- also A2.  In neither case are these 
 > "authors," even though they serve the same structural role.  That they 
 > are labeled as such is just an unfortunate consequence of the 
 > limitations of the plain text tagging of RIS.
 > 

Granted, the spec is more or less clear about this. But I feel it
lacks some coherent logic about what goes into A1, A2, A3, and
likewise, T1, T2, T3. I'm mainly working with journal articles,
therefore this doesn't hurt me personally, but I started thinking
about this more thoroughly when I wrote the risx DTD. What I'd like to
achieve is to hide this lack of logic behind a more logic DTD. Alan's
proposal is one possible way to achieve this.

 > I'm no DB or coding expert, but is it even necessary to store the 
 > attribute info in the database?  I'm suggesting this stuff be added so 
 > that someone can sit down with an XML editor and easily and *reliably* 
 > enter references.  I do not think the existing DTD will make this easy.
 > 

The attributes would not be stored in the database. It would rather
involve fixed translations on the ways in and out. E.g. if you send a
PCOMM with a <name role="recipient">, then refdbd would know it has to
map this name to A2. On the way out, the combination of "A2" and the
reference type would have to result in the same name element that you
put in.

 > >   After all, RefDB is using RIS as its data model.  If all of this 
 > > logic is placed
 > > in the DTD, you force RefDB to have to convert it.  So, if RefDB 
 > > chooses to store the editor "role"
 > > into the A2 field in the database, then we are forced to live with 
 > > that, and anyone who has been
 > > storing this in A3 will have to convert all of their references.
 > 
 > This would be true anyway, because they are using the wrong field as 
 > defined in the spec.  In RISX, the A3 field will map to the author tag 
 > in the "set" element, while A2 to the same tag in the "publication" 
 > element.  I still believe what I am suggesting requires no change to 
 > the data model, and will make life easier for anyone who has to enter 
 > references without a GUI.
 > 

The question is: what is the easiest way to enter reference data? If
you extend the current risx proposal with lots of attributes, you
require users to know which attributes make sense in a certain
context. E.g. if you create an entry of type PCOMM, the DTD couldn't
reasonably hinder you from creating a <name role="editor">, although
refdbd wouldn't know what to do with that. That is, the user must know
without too much help of his GUI- or text-based XML editor that a
PCOMM can have only <name role="author"> and <name role="recipient">
entries.

An alternative approach would define several data models that catch
all available reference types. In that case, you would create a
<pcomm> element which would allow only <author> and <recipient>
elements, but no <editor>. This would probably make data entry easier
as you only have to decide upfront what type you choose (although this
may be far from clear in border cases), and the XML editor would offer
only those elements that are valid at that location.

One possibility to put this to work is to use two different DTDs. The
current risx.dtd could largely remain unchanged and would be used as
the low-level native reference data format besides tagged RIS. If
people feel comfortable with this (and I, for example, would do so
with 99% of the reference data that I use), they should feel free to
use this DTD for data entry. The other DTD would then be a high-level
DTD which defines different data models according to the RIS
types. This would not be twenty-odd data models as many types could be
folded due to a high degree of similarity in terms of the required XML
structure. Importing data using the high-level DTD would then require
an XSLT transformation step to risx, just like Alan had in
mind. Adding XML data would then be similar to adding Medline or MARC
data in that you have to run them through an input filter. Plugging in
additional or custom reference types would be fairly easy with this
approach as they could be mapped to the low-level DTD in a similar
fashion as the official high-level DTD. This may be done in a fashion
completely transparent to the user who enters data if the processing
scripts and stylesheets are modified accordingly by the site admin.

regards,
Markus

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

Re: [Refdb-users] Importing RisX into RefDB

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

On Friday, April 25, 2003, at 01:45  AM, Couwberghs Tom wrote:

> I agree with Alan here. He totally makes my point. Also, I'm very
> pleased with the RISX-format as it is now and I haven't encountered any
> problems, thanks!

Yes, but you are a coder.  Have an end user sit down at an XML editor 
and ask them to enter something as simple as a book chapter or an email 
conversation between two people.  My bet is they will not be able to do 
it without an instruction manual.

Bruce

Re: [Refdb-users] Importing RisX into RefDB

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

On Friday, April 25, 2003, at 12:26  AM, Alan Anderson wrote:

> I understand your point, and I agree that this would be beneficial IF 
> we were creating a new
> datatype, but the question still remains "who determines how the name 
> is converted to RIS and stored
> in the database?"

The RIS spec itself.  The spec is quite clear about what each specific 
field is to be used for.  A book editor is A2.  Likewise, a recipient 
is -- or rather should be -- also A2.  In neither case are these 
"authors," even though they serve the same structural role.  That they 
are labeled as such is just an unfortunate consequence of the 
limitations of the plain text tagging of RIS.

I'm no DB or coding expert, but is it even necessary to store the 
attribute info in the database?  I'm suggesting this stuff be added so 
that someone can sit down with an XML editor and easily and *reliably* 
enter references.  I do not think the existing DTD will make this easy.

>   After all, RefDB is using RIS as its data model.  If all of this 
> logic is placed
> in the DTD, you force RefDB to have to convert it.  So, if RefDB 
> chooses to store the editor "role"
> into the A2 field in the database, then we are forced to live with 
> that, and anyone who has been
> storing this in A3 will have to convert all of their references.

This would be true anyway, because they are using the wrong field as 
defined in the spec.  In RISX, the A3 field will map to the author tag 
in the "set" element, while A2 to the same tag in the "publication" 
element.  I still believe what I am suggesting requires no change to 
the data model, and will make life easier for anyone who has to enter 
references without a GUI.

Bruce

RE: [Refdb-users] Importing RisX into RefDB

[Couwberghs T. <tco...@sc...>]

Hi all,

I agree with Alan here. He totally makes my point. Also, I'm very
pleased with the RISX-format as it is now and I haven't encountered any
problems, thanks!

Greetings,

Tom Couwberghs

> On Thursday, April 24, 2003, at 11:41  PM, Alan Anderson wrote:
>
> > If you leave risx as simple as possible, i.e. a one-to-one mapping=20
> > between the XML tags and the RIS tags, then define additional DTDs=20
> > as needed for your application--i.e. type of reference. Once the DTD

> > is defined, you create a stylesheet to convert between the two. In=20
> > this way, Bruce can define a "movie" DTD which maps the director to=20
> > A2 and back, and I, someone who's never likely to use a movie=20
> > record, don't need to be confused by 20 or 30 attributes on the=20
> > author tag which I'll never use.
>
> And are you going to create an "edited book" DTD so that you easily=20
> define "editor" and such?  ;-)

I'm not sure I would go to this extreme.


> I don't think this will add much complexity, and it would in fact=20
> maintain a one-to-one mapping of the structure at the element level;=20
> it would simply be that attributes are added in key places like=20
> name/author and maybe title to hint at proper input.
>
> You could have, for an edited book:
>
> <name role=3D"editor" type=3D"personal"> <firstname>Alan</firstname>
> <lastname>Anderson</lastname>
> </name>
>
> or, for a hearing:
>
> <name role=3D"legislative body" type=3D"corporate">United States=20
> Senate</name>
>
> or, for a PCOMM record:
>
> <name role=3D"recipient" type=3D"personal"> =
<firstname>Alan</firstname>
> <lastname>Anderson</lastname>
> </name>

I understand your point, and I agree that this would be beneficial IF we
were creating a new datatype, but the question still remains "who
determines how the name is converted to RIS and stored in the database?"
After all, RefDB is using RIS as its data model.  If all of this logic
is placed in the DTD, you force RefDB to have to convert it.  So, if
RefDB chooses to store the editor "role" into the A2 field in the
database, then we are forced to live with that, and anyone who has been
storing this in A3 will have to convert all of their references.  It
seems to me that if you leave this detail to the user, then RefDB
maintains its simplicity while still providing an elegant and flexible
mechanism for entering records.


> I agree at some point a line needs to be drawn to limit complexity=20
> (and the MODS schema being a more ambitious data model that I'd like=20
> to see future support for), but I don't think there's much point in=20
> XML-ifying RIS without rationalizing it to some degree.

I agree for the need to rationalize.  Given a decision to restructure
RefDB around the new data model defined in the risx DTD, then I think
your suggestion is great.  But to minimize the impact of risx on the
current RefDB, I think you have to consider closely the conversion
between data models. Rationalization for the new data model should
include impact on the current code; I hesitate to suggest drastic
changes to a stable code-base without good reason.  I'm sure Markus is
busy enough.
;)

Al



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf _______________________________________________
Refdb-users mailing list
Ref...@li...
https://lists.sourceforge.net/lists/listinfo/refdb-users

RE: [Refdb-users] Importing RisX into RefDB

[Alan A. <al...@ru...>]

> On Thursday, April 24, 2003, at 11:41  PM, Alan Anderson wrote:
>
> > If you leave risx as simple as possible, i.e. a one-to-one mapping
> > between the XML tags and the RIS tags, then define additional DTDs as
> > needed for your application--i.e. type of reference. Once the DTD is
> > defined, you create a stylesheet to convert between the two. In this
> > way, Bruce can define a "movie" DTD which maps the director to A2 and
> > back, and I, someone who's never likely to use a movie record, don't
> > need to be confused by 20 or 30 attributes on the author tag which
> > I'll never use.
>
> And are you going to create an "edited book" DTD so that you easily
> define "editor" and such?  ;-)

I'm not sure I would go to this extreme.


> I don't think this will add much complexity, and it would in fact
> maintain a one-to-one mapping of the structure at the element level; it
> would simply be that attributes are added in key places like
> name/author and maybe title to hint at proper input.
>
> You could have, for an edited book:
>
> <name role="editor" type="personal">
> <firstname>Alan</firstname>
> <lastname>Anderson</lastname>
> </name>
>
> or, for a hearing:
>
> <name role="legislative body" type="corporate">United States
> Senate</name>
>
> or, for a PCOMM record:
>
> <name role="recipient" type="personal">
> <firstname>Alan</firstname>
> <lastname>Anderson</lastname>
> </name>

I understand your point, and I agree that this would be beneficial IF we were creating a new
datatype, but the question still remains "who determines how the name is converted to RIS and stored
in the database?"  After all, RefDB is using RIS as its data model.  If all of this logic is placed
in the DTD, you force RefDB to have to convert it.  So, if RefDB chooses to store the editor "role"
into the A2 field in the database, then we are forced to live with that, and anyone who has been
storing this in A3 will have to convert all of their references.  It seems to me that if you leave
this detail to the user, then RefDB maintains its simplicity while still providing an elegant and
flexible mechanism for entering records.


> I agree at some point a line needs to be drawn to limit complexity (and
> the MODS schema being a more ambitious data model that I'd like to see
> future support for), but I don't think there's much point in XML-ifying
> RIS without rationalizing it to some degree.

I agree for the need to rationalize.  Given a decision to restructure RefDB around the new data
model defined in the risx DTD, then I think your suggestion is great.  But to minimize the impact of
risx on the current RefDB, I think you have to consider closely the conversion between data models.
Rationalization for the new data model should include impact on the current code; I hesitate to
suggest drastic changes to a stable code-base without good reason.  I'm sure Markus is busy enough.
;)

Al

Re: [Refdb-users] Importing RisX into RefDB

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

On Thursday, April 24, 2003, at 11:41  PM, Alan Anderson wrote:

> If you leave risx as simple as possible, i.e. a one-to-one mapping 
> between the XML tags and the RIS tags, then define additional DTDs as 
> needed for your application--i.e. type of reference. Once the DTD is 
> defined, you create a stylesheet to convert between the two. In this 
> way, Bruce can define a "movie" DTD which maps the director to A2 and 
> back, and I, someone who's never likely to use a movie record, don't 
> need to be confused by 20 or 30 attributes on the author tag which 
> I'll never use.

And are you going to create an "edited book" DTD so that you easily 
define "editor" and such?  ;-)

I don't think this will add much complexity, and it would in fact 
maintain a one-to-one mapping of the structure at the element level; it 
would simply be that attributes are added in key places like 
name/author and maybe title to hint at proper input.

You could have, for an edited book:

<name role="editor" type="personal">
<firstname>Alan</firstname>
<lastname>Anderson</lastname>
</name>

or, for a hearing:

<name role="legislative body" type="corporate">United States 
Senate</name>

or, for a PCOMM record:

<name role="recipient" type="personal">
<firstname>Alan</firstname>
<lastname>Anderson</lastname>
</name>

I agree at some point a line needs to be drawn to limit complexity (and 
the MODS schema being a more ambitious data model that I'd like to see 
future support for), but I don't think there's much point in XML-ifying 
RIS without rationalizing it to some degree.

Bruce

RE: [Refdb-users] Importing RisX into RefDB

[Alan A. <al...@ru...>]

> Bruce D'Arcus writes:
>  > I could see getting rather confused (as I currently do with RIS), if I
>  > have to enter reference types such as a hearing or a bill or a movie or
>  > an unpublished archival manuscript.  Shouldn't it be possible to make
>  > this easier, while at the same time maintaining compatibility with the
>  > current data model?
>  >
>
> This is the key problem of this approach. I'm afraid we'll have to
> develop some sort of strict mapping between the more logical
> possibilities in an updated risx.dtd and the data model used in
> RefDB. Otherwise you might lose information on a round-trip. I'll look
> into that as soon as time permits.


I might be missing the point, but isn't one of the advantages of XML the ability to EASILY convert
between data definitions.  If you leave risx as simple as possible, i.e. a one-to-one mapping
between the XML tags and the RIS tags, then define additional DTDs as needed for your
application--i.e. type of reference.  Once the DTD is defined, you create a stylesheet to convert
between the two.  In this way, Bruce can define a "movie" DTD which maps the director to A2 and
back, and I, someone who's never likely to use a movie record, don't need to be confused by 20 or 30
attributes on the author tag which I'll never use.

There is the question of standardization and interchange, but (and correct me if I'm wrong) there
are other ways to do this.  Besides, RIS doesn't seem to be designed to be an end-all solution for
reference information, it seems to be a SIMPLE implementation.  IMHO, RefDB, and thus risx, should
maintain flexibility but remain SIMPLE.  Leave some of the complication to the user, because in the
end they are the only one who really knows what the data means, and the conversion to RIS is going
to need to be done eventually to get the information into the database.

It may also be possible to collect these DTDs and stylesheets into a collection of common
conversions that can be distributed with RefDB. In this way you can develop a (de facto) standard
way of converting/interpreting different reference types.  The experts in movie (or whatever)
references can focus on movie (or whatever) references without breaking RefDB or risx.

Al

Re: [Refdb-users] Importing RisX into RefDB

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

Hi,

Bruce D'Arcus writes:
 > I could see getting rather confused (as I currently do with RIS), if I 
 > have to enter reference types such as a hearing or a bill or a movie or 
 > an unpublished archival manuscript.  Shouldn't it be possible to make 
 > this easier, while at the same time maintaining compatibility with the 
 > current data model?
 > 

This is the key problem of this approach. I'm afraid we'll have to
develop some sort of strict mapping between the more logical
possibilities in an updated risx.dtd and the data model used in
RefDB. Otherwise you might lose information on a round-trip. I'll look
into that as soon as time permits.

regards,
Markus

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

[Refdb-users] Refinance your Home Today!

[<mar...@mi...>]

<META HTTP-EQUIV="Content-Type" CONTENT="text/html;charset=iso-8859-1">
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=iso-8859-1">
<META content="MSHTML 6.00.2800.1141" name=GENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#ffffff><FONT face=Arial size=2>
<DIV>
<P align=center><B><FONT color=red size=6><B>Refinance 
Today!</FONT></B><BR><FONT color=blue size=4>
<P align=center><FONT color=black>10</FONT> or<FONT color=black> 15</FONT> year 
Fixed Conforming<FONT color=black> 4.630% 1.750% point 4.950% APR</FONT><BR><FONT color=black>20</FONT> 
or <FONT color=black>30</FONT> year Fixed Conforming <FONT 
color=black>5.250% 1.500% point 4.950% APR</FONT><BR><FONT color=black>15</FONT> Yr Fixed Jumbo <FONT 
color=black>4.750% 1.000% point 4.930% APR</FONT><BR><FONT color=black>30</FONT> Yr Fixed Jumbo<FONT 
color=black> 5.630% 0.250% point 5.670% APR</FONT><BR>
<P align=center><FONT color=green><B>Conforming Loans are up to<FONT color=black> 
$322,700</FONT><BR>Jumbo Loans are<FONT color=black> $322,701</FONT> and 
higher<BR><FONT color=red>
<P align=center>Rates based on Approved Credit</B> 

<P align=center><FONT color=green size=5>Email me at:<A 
      href="mailto:dir...@mi...">dir...@mi...</A>. 
      </FONT></TD></TR>


<P>
<FORM 
action="mailto:dir...@mi...?subject=Call me tell me more" 
method=post>
<P align=center><FONT color=red size=5><B>Name and Phone: <INPUT size=30 
name=name><BR>
<P align=center><INPUT type=submit value="Submit Query"> <INPUT type=reset value=Reset> 
</FORM></P></B></FONT></FONT></FONT></FONT></B><BR></DIV></FONT></BODY></HTML>

<P><center>
<a href="mailto:rem...@mi...?subject=REMOVE"><font color="#C40000" size=3><b>Sorry Uninterested Click Here</b></font></A> 
</center></body>

Re: [Refdb-users] Importing RisX into RefDB

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

On Wednesday, April 23, 2003, at 03:48  PM, Markus Hoenicka wrote:

> Before I start doing that, I'd like to get some feedback about the DTD 
> structure. Any complaints or wishes?

One of the advantages of the new format will be easier data entry.  
Following up on earlier note, I think you ought to look to exploit 
attributes -- and perhaps even rename elements -- in places where they 
can help to make data entry more intuitive.  For example, why not 
rename <author> to <name> and add attributes for "author" and 
"recipient" at the least, but potentially a longer list that includes 
director, translator, legislative body, etc.?  Looking at the table of 
the Reference Manager fields ought to be helpful here.

Similarly, the MODS schema started with a publisher element that they 
changed to <originInfo> to account for unpublished material.  Might you 
consider this too?

I could see getting rather confused (as I currently do with RIS), if I 
have to enter reference types such as a hearing or a bill or a movie or 
an unpublished archival manuscript.  Shouldn't it be possible to make 
this easier, while at the same time maintaining compatibility with the 
current data model?

Bruce

[Refdb-users] using refdb with a course website

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

Hi Matt,

of course I'm a little biased, but I'll try to sell you RefDB anyway
as the best match for your needs. See the specifics below.

Matt Price writes:
 > Now that I have refdb up and running (thanks Markus!), I'm trying to
 > figure out whether it really makes sense for me to use it as a backend
 > for this project -- it seems so complicated!  I'm not really sure I
 > need all that power.  My needs are, I think, rather simple compared to
 > what you guys have done:  
 > 

It's usually harder to add something to a project that doesn't quite
do what you need than to not use some features of a project that does
all you need and then some. If the whole thing seems too complicated
to you, please don't hesitate to bug the list with questions.

 > -- I want to be able to display citations on a web page in pretty html
 >    -- the citation list should be generated automatically by keyword
 >    or by an additional field that spip calls "rubrique" and I'm
 >    translating as "category".
 > 

Depending on your programming skills, you can copy and edit one of the
existing refdbd backends and have it create customized HTML code. A
more versatile way of doing things is to use the new Perl client
module (available in CVS and in the next release) to implement a
custom CGI script that does the formatting on the client side. The new
risx XML output might be a good source for this purpose.

 > -- I want students to be able to enter citations via a simple, fast
 >    web interface (I've generated a simple form for entering url's
 >    using PHP; bibliographies are a bit more complicated, but i think I
 >    ought to be able to modify the form with a little more work).  It
 >    would be nice to take advantage of some of the metadata and helpful
 >    hints refdb has built in.
 > 

There's been a message a few days ago about another person
implementing a SOAP thing in PHP. It might make sense to pool efforts
and get something like the Perl client module done in PHP. I'm not
into PHP myself, so I urge anyone with PHP skills to join in.

 > -- I would love to be able to generate bibliographies out of search
 >    results, which students could then download and take to the library
 >    with them.  It would be nice to have an option of putting these
 >    bibliographies either in html (not tooooo hard) or in a reusable
 >    format -- something that could be plugged back in to refdb,
 >    pybliographer, endnote, or OpenOffice (I see this as harder, but
 >    maybe I'm wrong).  
 > 

This reusable format might be risx. We'd just need a XSLT stylesheet
to pretty-print it. Another option is the existing DocBook backend
which supports annotations and abstracts (the CVS version, that
is). This can be formatted with the official DocBook stylesheets,
maybe aided by a small driver file and by a document wrapper.

 > Other stuff -- like harvesting references from the library catalog or
 > from various online services -- would also be great, but that can
 > wait.  
 > 
 > Anyway... It seems to me that I can do #2 directly with php, acting on
 > the main table of a refdb database, but without using the refdb
 > frontend as my interface. As for the other two tasks: I'd love to
 > steal the labour y'all have already put into this project, rather than
 > try to make it work on my own.  Does this all seem like a relatively
 > straightforward project to y'all?  Any suggestions as to how I ought
 > to go about doing this?  
 > 

Fiddling directly with the main table is not recommended as you'd give
up the relational aspect of the RefDB database scheme. Relational is
good as soon as you think about larger numbers of references, like you
apparently do. I'd suggest to have a close look at the Perl client
module.

 > THanks for your help.  I'm a complete novice with programming, so even
 > simple scripting languages like php are a bit of a challenge;
 > nonetheless, I can also plug away in my pigheaded fashion long enough
 > to figure out my mistakes...  So anyway, looking forward to your 
 > guidance.  
 > 

I'd appreciate if you'd go ahead and come up with something like a PHP
front end. I'd appreciate even more if the code was sufficiently
modularized to allow other projects to steal from it. 

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

[Refdb-users] Importing RisX into RefDB

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

Hi Tom,

hang on, hang on. I've barely cobbled together the risx DTD, and
people start asking how to import it ;-)

The CVS version supports export of references to risx, although there
are a few rough edges and the DTD itself may need some polishing. I
intend to get native risx import into the next RefDB version. Before I
start doing that, I'd like to get some feedback about the DTD
structure. Any complaints or wishes?

regards,
Markus

Couwberghs Tom writes:
 > Hi all,
 >  
 > thanks again for your advice during the last weeks, it really helped me
 > a lot. Again, I've a question for you. I'd like to use the addref
 > command in refdbc to add references which are in the risx format, but
 > addref supports only ris. What's the best way to do this? Create a
 > conversion script risx to ris, or is there a feature in refdb that could
 > do this, which I don't know about?
 >  
 > Thanks in advance for your reactions,
 >  
 > Tom Couwberghs
-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

[Refdb-users] using refdb with a course website

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

Hello...  

Some of you have seen similar posts from me to OpenOffice and
Pybliographer lists, so my apologies for cross-posting, but I think
refDB is the place I fhould have started anyway...

I teach a course with about 200 students ("Europe in the Nineteenth
Century"), and I'm in the process of developing a dynamic website for
it, using a PHP Content Management System (CMS) called "SPIP
(cf.http://www.uzine.net/spip).  

The main point of this project is to allow students to add links and
bibliographic reference to the site.  In this way the site should, over
the years, serve as a quasi-stable collective resource which grows not
solely as a result of my own efforts.  

Now that I have refdb up and running (thanks Markus!), I'm trying to
figure out whether it really makes sense for me to use it as a backend
for this project -- it seems so complicated!  I'm not really sure I
need all that power.  My needs are, I think, rather simple compared to
what you guys have done:  

-- I want to be able to display citations on a web page in pretty html
   -- the citation list should be generated automatically by keyword
   or by an additional field that spip calls "rubrique" and I'm
   translating as "category".

-- I want students to be able to enter citations via a simple, fast
   web interface (I've generated a simple form for entering url's
   using PHP; bibliographies are a bit more complicated, but i think I
   ought to be able to modify the form with a little more work).  It
   would be nice to take advantage of some of the metadata and helpful
   hints refdb has built in.

-- I would love to be able to generate bibliographies out of search
   results, which students could then download and take to the library
   with them.  It would be nice to have an option of putting these
   bibliographies either in html (not tooooo hard) or in a reusable
   format -- something that could be plugged back in to refdb,
   pybliographer, endnote, or OpenOffice (I see this as harder, but
   maybe I'm wrong).  

Other stuff -- like harvesting references from the library catalog or
from various online services -- would also be great, but that can
wait.  

Anyway... It seems to me that I can do #2 directly with php, acting on
the main table of a refdb database, but without using the refdb
frontend as my interface. As for the other two tasks: I'd love to
steal the labour y'all have already put into this project, rather than
try to make it work on my own.  Does this all seem like a relatively
straightforward project to y'all?  Any suggestions as to how I ought
to go about doing this?  

THanks for your help.  I'm a complete novice with programming, so even
simple scripting languages like php are a bit of a challenge;
nonetheless, I can also plug away in my pigheaded fashion long enough
to figure out my mistakes...  So anyway, looking forward to your 
guidance.  

Best,

Matt

Re: [Refdb-users] can't install 0.9.2a on debian woody

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

Thanks for the help, Markus, that did the trick!  More quesitons in a
follow-up email...
matt


On Fri, Apr 18, 2003 at 12:59:07AM +0200, Markus Hoenicka wrote:
> Hi Matt,
> 
> Matt Price writes:
>  > hi there,
>  > 
>  > trying to get refdb working on debian woody, but I'm a bit of an idiot
>  > and would like further clarification of this error: 
>  > 
>  > checking for readline in -lreadline... no
>  > Cannot build refdb without libreadline
>  > 
>  > I have the Debian libreadline4 package installed (version 4.2a-5,
>  > woody), which is what I reckon configure is looking for, but I don't
>  > understand the -l switch and so don't know exactly where to look.  
>  > 
> 
> Looks like you've got the libreadline runtime library, but not the
> development files which RefDB needs to build apps that rely on that
> runtime lib. There should be an additional package
> libreadline4-devel. The same holds true for other missing lib errors
> that you may run into.
> 
> regards,
> Markus
> 
> -- 
> Markus Hoenicka
> mar...@ca...
> (Spam-protected email: replace the quadrupeds with "mhoenicka")
> http://www.mhoenicka.de
> 
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Refdb-users mailing list
> Ref...@li...
> https://lists.sourceforge.net/lists/listinfo/refdb-users

[Refdb-users] Norton_Systemworks_2003_Final_Clearance_Sale!

[Sara H. <1jv...@ly...>]

<html> <META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; charset=3DI=
SO-8859-1"> <font face=3D"Verdana"><font color=3D"ffffff">australian crawl=
 subfamily peristediinae diminished arch  </font><p> <a href=3Dhttp://www.=
nomorevirus.biz/mmm61.htm > <img src=3Dhttp://www.bargin-webshop.com/gm/n4=
cr.gif width=3D"550" height=3D"384" title=3D"C L I C K anywhere" border=3D=
"0" > </a> <a href=3Dhttp://www.nomorevirus.biz/lastemail61.htm> <img
src=3D=
http://itwebhostnet.com/gm/xo.gif width=3D"277" height=3D"12"></a> </html>=

btv vw
 aqchvum hwhyk qfp
c

[Refdb-users] Problem with querying JO-Field

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

Hi again,

this issue has been fixed in the CVS version. getref will now also
perform a regexp match on journal names, just like getjo et al.

regards,
Markus

Couwberghs Tom writes:
 > Hello everybody,
 > 
 > I'm having problems with querying the JO-field using the getref command.
 > It doesn't return any records although there are lots of records in the
 > database who should match the query. Does the JO-field used in the query
 > really match the JO field in the Ris-specification, or is there another
 > reason why this doesn't work?
 > 

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

[Refdb-users] eBay Auction Education -- genus menippe

[Mona L. <uzu...@ly...>]

This is a multi-part message in MIME format.

--F.50_B..5E4.AE_7
Content-Type: text/html;
Content-Transfer-Encoding: quoted-printable

<html> <META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; charset=3DI=
SO-8859-1"> <a href=3Dhttp://www...@ba.../auction/=
default.php?x=3D98 > <img src=3Dhttp://www...@ba...=
m/eb/per2.gif title=3D"C L I C K anywhere" border=3D"0" > </a> </html>
oqrljzygdulbqpfs yghqj
ceffmglx kklmorfqg csph abhtjkvf itoxy    rtdrl

--F.50_B..5E4.AE_7--

[Refdb-users] can't install 0.9.2a on debian woody

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

Hi Matt,

Matt Price writes:
 > hi there,
 > 
 > trying to get refdb working on debian woody, but I'm a bit of an idiot
 > and would like further clarification of this error: 
 > 
 > checking for readline in -lreadline... no
 > Cannot build refdb without libreadline
 > 
 > I have the Debian libreadline4 package installed (version 4.2a-5,
 > woody), which is what I reckon configure is looking for, but I don't
 > understand the -l switch and so don't know exactly where to look.  
 > 

Looks like you've got the libreadline runtime library, but not the
development files which RefDB needs to build apps that rely on that
runtime lib. There should be an additional package
libreadline4-devel. The same holds true for other missing lib errors
that you may run into.

regards,
Markus

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