To answer the call for a few ideas and observations:
I would like to see a next generation catalog that has a dedicated
display for each entity, as defined by FRBR, but not limited to FRBR.
The design idea that I envision so far is based on LibraryThing's "link
to a page, not to a search" philosophy and on OCLC's WorldCat Identities
http://orlabs.oclc.org/Identities/.
A main reason for that type of display comes from Working on the
displays and indexes for our Horizon-based system over the years.
I broke down the display issues into three categories:
1. browse list displays
2. bib summaries or search result displays (from keyword searches, or
results after selecting headings from a list)
3. full bib displays, or MARC record displays with attached holdings
information
Over the years there have been enhancements to these types of displays,
in our system and in other systems.
1. Browse lists. Not many enhancements have appeared over the years, but
I have seen authority information such as scope notes appear in pop-up
options on other systems. Idea for further enhancement: use BT, NT, RT
for subject browses, instead of just SEE or SEE ALSO.
2. Search result screens. The new thing is the facet and cluster options
from providers like Endeca.
Other enhancements, such asnriched content like cover art and holdings
information under each hit, have been added over the years. It seems
there's been a lot of effort to re-create the Google/Amazon feel at this
level, but to add library know-how with more accurate and precise
results.
3. With full bib display I've seen many debates such as the use of
labels instead of ISBD-based displays. In recent years enriched content
such as cover art, reviews, and table of contents have been added.
Holdings information usually appears on the same screen, and in some
cases this has been enhanced with links to maps and links to call number
browse lists (creating a virtual shelf browse experience).
Hyperlinks in the full bib display tend to dive back into any of the
three types of displays, but what I think is missing here is a sense of
anchoring the user to the concept they have clicked on. I have selected
a record-- how does this result help me discover other/better/similar
resources? The challenge has been partly answered by the FRBR model and
also by social networking tools. After searching Amazon, people get
anchored on the final screen for the record of the book. At that point,
people can add reviews and tag and discuss and in so doing help other
users in the discovery process. With FRBR, I see a return to the
original principles of catalog design with a focus on collocation-- find
ALL works by an author, find ALL related editions of a work, find ALL
works on a topic. Clicking on an authority controlled heading from a
record does that to an extent, but many online catalogs don't handle
name-title headings well, so in many cases not ALL works are correctly
displayed and collocated.
Does MARC need to change?
One FRBRized ILS that hasn't been mentioned much is VTLS's Virtua
system. From what I've seen, MARC records are "FRBRized" by some sort of
decomposition into superwork, work, expression, and manifestation level
records, linked by customized 004 MARC tags. The resulting tree-like
structure of FRBRized records I don't find particularly user-friendly,
and so the great challenge I think is to design displays that can take
an idealized FRBRized data set and show the attributes and relationships
in ways that are consistent and user-friendly. I think many of the
problems with AACR2 and MARC stem from the underlying assumption that a
catalog card will be the final display option. The next generation
catalog should incorporate backwards compatibility as much as possible,
but we need to continue forward and think about display decisions that
are not hampered by card catalog design limits.
A few years ago my library upgraded to a version of Horizon that used
the feature of multi-user authorities. Previously each index had its own
authorities. A problem arose when, for example, an authority record for
"William Shakespeare" was updated in the author index, but "William
Shakespeare" in the subject index was not affected. With multi-use
authorities, a change in one index is automatically picked up in the
other because there is only one underlying authority record. In FRBR
terms, there is only one record for the entity of "William Shakespeare"
and that record is then used as the basis of relationships with other
entities.
How do multi-use authorities help the user? This reminds me of the
distinction made at the meeting of the Working Group on the Future of
Bibliographic Control at the Library of Congress between consumer
users and management users. Multi-use authorities primarily seem to help
me in saving time editing records. They help the user a bit by making
sure headings are consistent. But suppose, as a catalog user, I was
interested in Shakespeare-- works by him, works about him, even works
related to his works.
Browsing an author index doesn't help me when I also want works about
him. I can see why keyword searches are so popular-- just punch in
"William Shakespeare" and you get everything (and more!). But the
results aren't differentiated properly. The faceted or clustered options
to control the scope of the result list help, but I think there's
something missing here.
When I look at FRBR, multi-user authorities, and some of the newer
display designs for catalogs, what I see as a possibility is the option
to look at a result that focuses on the entity of interest, and from
that focused display I see browse lists and search results added in. I'd
like to see a "William Shakespeare" web page as a result in a catalog
search with all relationships to all other entities mapped out on that
page. I'd like to see that web page serve as an anchor for social
networking tools so the discovery process can be enhanced more. I'd like
to see the idea of enriched content (cover art, reviews, etc.) brought
into and merged with the catalog data on that web page. I'd like to see
a "Did you really mean" feature for similar looking names.
I recently worked on a cataloging issue that seemed related to the
issues and opportunities I have been thinking about.
The current authority for the singer "Cat Stevens" is "Yusuf Islam". If
I search either name in a library catalog I get a listing of all names
with a SEE reference in the index. There were quibbles about "Yusuf
Islam" has the chosen form, but in the traditional library catalog way
of doing things, all works are collocated under that heading.
I checked how Amazon handles this.
I searched Amazon for "Yusuf Islam". The result indicated there was a
related heading: "Cat Stevens".
All good so far.
I selected a record that had "Yusuf Islam" as a hyperlink and I ran into
a problem. Clicking on a heading for "Yusuf Islam" in a specific record
for, say, "Footsteps in the Light" generates a search on the keywords
"Yusuf" and "Islam" only-- no sign of Cat Stevens anymore in the result
list. The social networking data (reviews, tags, other purchases) did
point to Cat Stevens. OK, so the value of social networking data is to
help fill the gaps in the raw keyword based approach, but wouldn't a
better solution be to focus on the entity first, and then have the
clarifications and connections and context flow from that? It seems
that's what FRBR is about. Wouldn't it be great to have a result with a
page for Cat Stevens/Yusuf Islam, that would list all of his works,
would have links to added value content and social networking data, and
have the smarts to differentiate between works by him and works about
him?
It seems that a lot of the work on FRBR (and perhaps now RDA/DCAM/RDF?)
point to that, but we might be hampered by our visions of the past,
rooted in card catalog designs and the need to continue to support that
common denominator. Perhaps we should start with visions of what we want
and work backwards from there.
Thomas Brenndorfer, B.A, M.L.I.S.
Guelph Public Library
100 Norfolk St.
Guelph, ON
N1H 4J6
(519) 824-6220 ext. 276
tbrenndorfer_at_library.guelph.on.ca
Received on Thu May 10 2007 - 10:22:38 EDT