ISKO 2010 Conference , February 23-26, 2010
Sapienza University of Rome, Italy
Locally Added Homegrown Metadata Semantics: Issues and Implications
Jung-ran ParkYuji TosakaCaimei LuResearch supported through IMLS award (2006-2010)
22
Presenters
Jung-ran Park, Ph.D.Assistant Professor, College of Information Science and Technology, Drexel University, Philadelphia, Pennsylvania, USA
Yuji Tosaka, Ph.D.Cataloging/Metadata LibrarianThe College of New Jersey Library, Ewing, New Jersey, USA
Caimei LuPh.D. candidate, College of Information Science and Technology, Drexel University, Philadelphia, Pennsylvania, USA
2
33
Presentation outline
•Aim and scope of the study•Previous studies•Data collection & methods•Study findings•Conclusion/implication
3
44
Locally added metadata elements•Proliferation of local metadata extensions and
variants to accommodate domain-specific requirements▫Benefit: Provide rich, detailed descriptions for local
communities ▫Downside: Roadblock to metadata interoperability
5
Objectives of the study•To gain a better understanding of the extent of the
current use of locally added metadata elements and their potential effects on resource discovery and sharing across digital repositories.
5
66
Study questions•What homegrown metadata elements are added in
individual digital repositories?•What are the major criteria used to add locally
created homegrown metadata elements?•How are local metadata practices documented and
shared?
77
Previous studies• “Metadata may be of high quality within its local
context, but may be compromised when taken out of this context for various reasons.” (Shreeves, Riley, & Milewicz, 2006)
• Little empirical research on how local metadata elements are added across distributed digital repositories▫ Park (2005), Han et al. (2009)
8
Previous studies: Park (2005)•Analysis of 659 metadata records in three digital
image collections using Dublin Core▫Widespread use of locally added metadata elements ▫ Some of the most frequent additions are local
information such as “contact information,” “ordering information,” and “acquisition.”
▫ Presence of null mapping local elements such as “full text,” “note,” and “scan date,” although they can be mapped onto pertinent DC metadata elements.
8
9
Previous studies: Han et al. (2009)•Analysis of locally defined unique fields created by
21 CONTENTdm-based collections using Dublin Core •107 (84.3%) out of 127 unique descriptive metadata
fields can be mapped onto pertinent DC metadata elements. ▫ DC metadata semantics seems to affect the correct
application of the DC metadata scheme in local contexts (see Park and Childress, 2009).
9
1010
Data collection & methods (1)•Web Survey (August 6–October 6, 2008)•Participants recruitment▫Mailing lists of communities of metadata and
cataloging professionals▫Individual invitations▫Flyers distributed at selected metadata and cataloging
sessions during the 2008 ALA midwinter conference
1111
Data collection & methods (2)•303 completed responses•Participant profiles▫Most survey respondents engage professionally in
cataloging- and metadata-related activities.• Institutional profiles▫Academic libraries (75.2%)▫Public libraries (17.4%) ▫Other institutions (7.4%)
12
Participant profiles
Job TitlesNumber of
Participants (N=227)
Other 135 44.6%Cataloger/cataloging librarian/catalog librarian 99 32.7%Metadata librarian 29 9.6%Catalog & metadata librarian 26 8.6%Head, cataloging 26 8.6%Electronic resources cataloger 17 5.6%Cataloging coordinator 15 5.0%Head, cataloging & metadata services 15 5.0%
13
Professional Activities Specified in “Other” Category
Professional Activities Number of Participants (N=99)
Cataloging & metadata creation 31 10.2%Digital projects management 23 7.6%Technical services 17 5.6%Archiving 16 5.3%Electronic resources & serials management
6 2.0%
Library system AdministrationOther (e.g., education-library and information science)
6 2.0%
1414
Current status of homegrown metadata element use•Common practice in many digital repositories▫38.0% add homegrown elements to the selected
metadata scheme(s)
15
Metadata schemes used
15
Metadata scheme Percent
MARC 84.2%
Dublin Core 66.0%
EAD 31.7%
MODS 17.8%
VRA 14.9%
TEI 12.5%
1616
Types of homegrown metadata elements added•Descriptive metadata (58.7%)▫Local notes and description (23.1%)▫Local names for person and place (22.1%)▫Local subjects (14.3%)▫Local classification (6.0%)▫Genre and type terms (4.5%)
•Administrative, technical, and preservation metadata (31.6%)
•Structural metadata (3.8%)
1717
Criteria for adding homegrown metadata elements
Response
To reflect the nature of local collections/resources 32.0%
To reflect the characteristics of target audience/community of local collections 24.3%
To reflect local practice of cataloging and metadata creation 14.3%
Due to lack of metadata description for local collections/resources from the selected metadata scheme(s) 14.3%
Due to local systems 7.7%
Don’t know 1.9%
Due to lack of full understanding of the selected metadata standard for local collections/resources 1.2%
Other 4.2%
1818
Current status of homegrown metadata element documentation practices•Are local extensions and variants allowed in local
metadata guidelines?▫Yes — 69.5%▫No — 14.9%▫Do not know — 15.6%
1919
Lack of open documentation•Only 19.6% of local documentation/guidelines for
metadata creation are publicly available on the Web.
2020
Conclusion/implication (1)• “Inward focus” in current metadata creation
practices▫Widespread use of homegrown metadata elements
(nearly 40%) ▫ Does the current cost of local metadata customization
make sense in light of its potential risk to effective sharing and reuse of metadata in the networked environment?
21
Conclusion/implication (2)•Most of the homegrown descriptive elements
(58.7%) are questionable as truly homegrown metadata elements because they can be replaced with relevant elements (e.g., description, subject) drawn from metadata schemes in use.
21
22
Conclusion/implication (3)•The high occurrence of locally added administrative
metadata elements encompassing technical, rights, and preservation metadata indicates that there is a need for metadata schemes to provide a more robust supporting mechanism and guidance for implementing administrative metadata elements in local contexts.
22
23
Conclusion/implication (4)•Domain-specific controlled vocabularies such as
genre and type terms for non-textual resources need to be further developed.
• Importance of application profiles that facilitate the meeting of local needs through adoption of metadata elements from various metadata schemes.
•Methods for facilitating the sharing of such local application profiles in turn would lead to increased metadata interoperability across digital repositories and collections.
23
24
Future studies•Rapidly growing body of digital repositories and
collections calls for further investigation of documentation practices (e.g., metadata best practices, metadata guidelines, application profiles).
•Novel approaches and techniques are needed for extracting, analyzing, and comparing those locally developed documentations.
24
25
Thank you!•Address post-conference correspondence to Jung-
ran Park ([email protected])
Jung-ran ParkYuji TosakaCaimei Lu