Transcript Collection Building Interfaces with Luna Insight

Collection Building Interfaces
with Luna Insight
Gale Halpern ([email protected])
Representing the Luna Development Team
Mira Basara, Rick Silterra, Surinder Ghangas
Growing Image Collections
Large dynamic image collections managed in Luna Insight
1.
2.
Herbert F. Johnson Museum of Art digitization project (Museum on-line) –
began in 1998.
Knight Visual Resources Facility digital image collection for instruction
within the Cornell Art, Architecture and Planning departments (Slide Library
on-line) – began in August 2007.
Smaller dynamic collections in Luna
3.
4.
Rare Books and Manuscript Digital Collection
New York Aerial Photographs
Luna
• has an ‘open’ architecture, allowing image collections to
interface to collection-specific ‘source’ tables.
• permits any collection-specific metadata schema which
can be mapped to industry-wide standards.
• is a digital delivery platform, not a repository. An
interface could be built between Luna and an institutional
repository.
Collection Sizes
Number of
Digital Images
(October 2007)
Anticipated Total
number of
Images
Current Image
Rate of growth
Herbert F. Johnson
Art Museum
collection
21,339
36,000 +
100 per month
Knight Visual
Resources
Collection
16, 359
unlimited
600 per month
Types of Collections
Herbert F. Johnson
Art Museum
collection
Knight Visual
Resources
Collection
Image Content
(mainly)
Maximum
Viewable Image
Resolution
Museum
Objects
24,576 pixels
(Permanent
Collection)
(lengthwise)
Scans of books,
1,536 pixels
slides, other
(lengthwise)
sources used for
instruction.
Copyright
Public Domain
except post-1923
(restricted)
Restricted
Different Challenges faced
• Where is the source data?
• platform (Oracle, Access,)
• commercial vs. homegrown software
• Metadata schema (Dublin-Core-like vs. VRA-like (Visual Resource
Assoc.))
• Data mapping between Luna and the feeder system
• Workflow/coordination of manual and automated tasks
• Frequency of update (once per month vs. once per week)
• Data quality – whose responsibility is it?
Workflow
How Luna collections are created?
• Metadata is catalogued by end-users.
• Images are scanned from slides/books or
objects photographed, then .tiffs are sent
to DCAPs for processing (to build .jpeg
derivatives).
• Data and Images are indexed and linked.
KVRF/Luna interface
Library 24 Server
Knight Visual Resources Facility Server
TEXT FILES
Works, Images,
Creators, Work
Relationships
PicTor Access
Database
Data Clean-up
(PERL scripts)
Scanned Images
(.tiffs)
Luna data upload
DCAPS
PC with Luna Media Batch Tools
Create
Derivatives
Image
Derivatives
(.jpegs)
Luna Insight
Oracle
Database
CD’s containing .tiffs
Luna Indexer
The Museum System(TMS)/Luna interface
Bonanzap Server (CIT)
Library 24 Server (DLIT)
Oracle DB Link
TMS Oracle
Database
Luna Insight
Oracle
Database
Luna Indexer
Photo Studio Server (Johnson Art Museum)
Digital Images
(.tiffs)
DCAPS
PC with Luna Media Batch Tools
Create
Derivatives
CD’s containing.tiffs
Image
Derivatives
(.jpegs)
PicTor
Knight Visual Resource Collection
Text Files
Works.txt
Images.txt
Knight Visual Resource
Collection
Data Compliance
• Built PERL scripts which reconcile problems in the data
– Normalize non-relational data
– Consolidate data stored in redundant locations
– Populate fields for Images with no Work Number
– Ensure correct display sequence (i.e. multiple titles,
creators, etc.)
Knight Visual Resource
Collection
Knight Visual Resource Collection
Interface – SQL View
• SQL view selects data
from the ‘cleaned up’
text file data.
• transforms flat Pictor
data to a normalized,
VRA-like format.
VRA is a Visual Resource Association
metadata standard
Knight Visual Resource Collection
Knight Visual Resource Collection
The Museum System (TMS)
Herbert F. Johnson Museum Collection
Part 1. TMS Database – SQL View
• TMS data structure is
proprietary & non-compliant
• View transforms TMS data
to HFJ compatible data
structure (Dublin Core-like)
• Created one TMS view per
HFJ DC-like table
Herbert F. Johnson Museum Collection
Part 2: Luna SQL View of a TMS SQL View
• hfj.bvtitle selects from vtitle @bonanzap
(the TMS server at CIT).
• Results of hfj.bvtitle are loaded into
hfj.bvt_table a table on the Luna server.
• Luna indexer runs against the
hfj.bvt_table.
Herbert F. Johnson Museum Collection
Herbert F. Johnson Museum Collection
What’s important for future?
Building future library systems:
• Buying/contracting for external solutions or building
blocks(Luna Insight, Artstor, The Museum System)
• Use of SQL views to transform metadata and build
interface.
• Using building blocks and interfaces (glue) to create
working systems.
Some thoughts on the future
• Create image collection repositories while maintaining the ability to
build collections (should Luna source tables be Fedora
repositories?)
• Improve the building blocks (i.e. replace Pictor with an Oracle
solution).
• Improve the metadata (shouldn’t these all be OAI-PMH compatible?)
• Migrate to real-time interfaces without human intervention.