“…all the triple stores in serious use today use more that 3 columns to store the triples. Instead of triples, RDF atoms are now stored as 4-tuples, 5-tuples, 6-tuples or 7-tuples.

…

Is there anything harmful with the misnomerization of “triple”, enough for the community to try their best to start talking about “tuples”? I think there is. Linked Data is the best example of how a focus on the three-ness of triples can fool people into sub-optimal implementations. I heard this fear expressed several times during the conference, although not in those words. More than once, people expressed concern that once data had been extracted via SPARQL and gone into the Linked Data cloud, there was no way to determine where the data had come from, what its provenance was, or whether is could be trusted. He was absolutely correct- if the implementation was such that the raw triple was allowed to separate from its source. If there was a greater understanding of the un-three-ness of real rdf tuplestores, then implementers of linked data would be more careful not to obliterate the id information that could enable trust and provenance. I come away from the conference both excited by Linked Data and worried that the Linked Data promoters seemed to brush-off this concern.”

“What has been the sole territory of relational database technologies may soon be displaced by the use of RDF and the triple store. Moreover, because RDF is the common data model utilized by triple stores, it is possible to integrate data sets across different triple stores – across different RDF data providers. This integration is conveniently afforded by the URI and RDF as web standards and is a function foreign to the relational database domain. With the Web of Data, no longer is information isolated in individual inaccessible data silos, but instead is exposed in an open and interconnected environment – the web environment.”

“In many respects, the most important question for the library world in examining semantic web technologies is whether librarians can successfully transform their expertise in working with metadata into expertise in working with ontologies or models of knowledge. Whereas traditional library metadata has always been focused on helping humans find and make use of information, semantic web ontologies are focused on helping machines find and make use of information. Traditional library metadata is meant to be seen and acted on by humans, and as such has always been an uncomfortable match with relational database technology. Semantic web ontologies, in contrast, are meant to make metadata meaningful and actionable for machines. An ontology is thus a sort of computer program, and the effort of making an RDF schema is the first step of telling a computer how to process a type of information.

…

To some extent, the success of the semantic web in any domain is predicated on the successful embodiment of that domain’s knowledge in ontological code. Either coders need to learn the domain knowledge, or domain experts need to learn to code. People need to talk.”

The problem from my standpoint is that MODS has some really odd, library-specific, design choices that I don’t think map very well to the wider world. A central concept like mods:name, with mods:role as a child of that, really makes no sense, and conflicts with more common modeling you see in DC, FRBR ,etc.

It’s semantics are also really loose.

So you have to ask yourself, just how linked could a MODS view in RDF really be?

Hi Bruce,

Thanks for your comment on my blog post:

More on Motivation for Investment in Implementation of a MODS Ontology
http://www.chrisfrymann.com/2009/07/31/more-on-motivation-for-investment-in-implementation-of-a-mods-ontology

I pretty much agree with what you write in your first two sentences:

“The problem from my standpoint is that MODS
has some really odd, library-specific, design choices
that I don’t think map very well to the wider world.
A central concept like mods:name,
with mods:role as a child of that,
really makes no sense, and conflicts with
more common modeling you see in DC, FRBR ,etc.

It’s semantics are also really loose.”

I also think you do ask a very important question:

“So you have to ask yourself, just how linked could a MODS view in
RDF really be?”

As I mentioned in my blog post:

The primary motivation for my work on a MODS ontology is directed toward
identifying an acceptable strategy for migrating existing MARC based metadata
into a form more universally accessible to consumers and producers of
Linked Data.

I’m not really deeply familiar with a lot of the history of community
discussion in this area, so over the weekend I took a look at your
blog, listened to a “Library Geeks” podcast you participated in a few
years ago, and browsed around a few other related links.

I know that such a brief intro can’t do anyone justice, but I think I
have a bit of an idea as to where you’re coming from, and I want to
say first thing that I am not particularly a defender of either MARC
or MODS. I am however a “big” supporter/promoter of concepts relating
to RDF, linked data and the various components of the “Semantic Web
Layer Cake”. Also, though I primarily consider myself a software
developer, and my title is “Digital Library Architect”, I work in a
library where MARC-based cataloging is still the primary form of
metadata creation and MODS/METS are probably the dominant leading edge
standards.

In my two most recent blog posts I tried to provide more detail on my
local situation in which we have implemented a very crude RDF
manifestation of MODS, have entered on the order of 15 million triples
into a triplestore (AllegroGraph), and are successfully using SPARQL
queries to search and process that data. We also feel that we have
developed a pretty nice JavaScript/JSON/AJAX-based user interface
enabling users to discover and view resources.

We would like to share some of this with others, but we also
understand that indeed no one else would be likely to adopt our
particular mode of what ultimately does amount to expressing MARC in
RDF. That is what got me started in thinking that if there really was
a community accepted/acceptable MODS ontology that we could use, it
could make our system more attractive to others.

As it stands logically, I think it would be “more” attractive to
others. However, your question:

“Just how linked could a MODS view in RDF really be?”

is a very good one, and I don’t really know enough to answer that. My
main thought has been simply that a “better” RDF view of MODS than the
one we are currently working with would at least be significant
progress. The serious question you seem to be posing though is that
even with a formal ontology for MODS, that still might not provide
enough new usability for others to consider adopting it as a serious
“standard”.

I only recently learned about your “Bibliographic Ontology
Specification” and though I have not yet had a chance to look at it
deeply, I am impressed with the direction you seem to be heading.

My first basic question though is, What thoughts, if any, do you have
regarding a strategy for migrating MARC to Bibliographic Ontology? I
expect you must have had some discussion in that area already and
would appreciate any references or comments you might send.

Finally, and this is just a fairly small point that I didn’t know
where else to add but here at the end. One of the things I did make a
somewhat feeble attempt to address in my MODS ontology is a little
better approach to handling the problem you mention of “role” being a
child element of “mods:name”. I know it’s not much help, but one
small, although I expect still controversial, thing I suggest is that
the various hierarchically structured MODS elements be given their own
individual predicate identifiers (such as: mods:name.role.roleTerm,
mods:name.role.roleTerm.authority, etc.) which would at least make
such items more directly SPARQL queryable without having to explicitly
build a SPARQL query that mirrors the graph structure of the fact
that terms like “role” may in some cases only be identifiable as a
graph subelement of some parent term like “name”. These identifiers
could also be assigned (owl:sameas) variants in whatever manner seems
most helpful, as ultimately the basic point of a MODS ontology would
largely be (or at least it is for us) to to specify a fixed/standard
set of MODS predicates. This could help to deal with the fact that
different MODS identifiers may have somewhat different meanings
depending on where they appear in the MODS hierarchy.

Sincerely,

- Chris

Chris Frymann
Digital Library Architect
University of California San Diego Libraries

blog: http://chrisfrymann.com

Hi Chris,

On Mon, Aug 3, 2009 at 4:12 PM, Chris Frymann wrote:

…

> As it stands logically, I think it would be “more” attractive to
> others. However, your question:
>
> “Just how linked could a MODS view in RDF really be?”
>
> is a very good one, and I don’t really know enough to answer that. My
> main thought has been simply that a “better” RDF view of MODS than the
> one we are currently working with would at least be significant
> progress. The serious question you seem to be posing though is that
> even with a formal ontology for MODS, that still might not provide
> enough new usability for others to consider adopting it as a serious
> “standard”.

Yes. I don’t think people outside of the library world pay much
attention to library standards, which are often taken to be odd and
baroque.

> I only recently learned about your “Bibliographic Ontology
> Specification” and though I have not yet had a chance to look at it
> deeply, I am impressed with the direction you seem to be heading.
>
> My first basic question though is, What thoughts, if any, do you have
> regarding a strategy for migrating MARC to Bibliographic Ontology? I
> expect you must have had some discussion in that area already and
> would appreciate any references or comments you might send.

I’ve never paid much attention to MARC.But the people at Talis and the
LOC use BIBO, and some of that work has been based on conversion from
MARC. If you’re interested in finding out more, you could always post
a question to the BIBO list?

Bruce

Hi Bruce,

Thanks for your reply and suggestion. I’ll try to learn more about
BIBO and how LOC and Talis may be using it.

Again, much of my current focus is on exploring potential paths for
releasing/repurposing the significant body of information that is
essentially locked up in MARC/MODS.

By the way, would you mind if I posted the text of this email
conversation in my blog? It might be of interest to others. Don’t
hesitate to let me know if you prefer not.

- Chris

> Hi Bruce,
>
> Thanks for your reply and suggestion. I’ll try to learn more about
> BIBO and how LOC and Talis may be using it.
>
> Again, much of my current focus is on exploring potential paths for
> releasing/repurposing the significant body of information that is
> essentially locked up in MARC/MODS.

Right.

> By the way, would you mind if I posted the text of this email
> conversation in my blog?

No, I don’t mind.

Bruce