From Spectator to Annotator: Possibilities offered by User- generated Metadata for Digital Cultural...
-
Upload
stephanie-wiley -
Category
Documents
-
view
220 -
download
0
Transcript of From Spectator to Annotator: Possibilities offered by User- generated Metadata for Digital Cultural...
From Spectator to Annotator: Possibilities offered by User-
generated Metadata for Digital Cultural Heritage Collections
Seth van Hooland
Université Libre de Bruxelles
Metadata creation for image collections
• Retrieval of high level semantics within images relies entirely on human indexing
• Indexing of historical image collections is notoriously hard and extremely expensive
• Digital images are created on a large scale (>10.000) • No specifically trained staff for attributing metadata
within the institution on an intensive basis• => Highly problematic to have enough inhouse
ressources to index image collections
Possible solution: distributed indexing
• Development of web-based collection management software at the end of ‘90s
• Possibilies to distribute the access of the database to a larger number of indexers
• The process of cataloging and indexing is no longer necessary an inhouse activity
• Example: http://na.memorix.nl/
Distributed image indexing: web2.0 tools
• Passive consumer of information => active user who reorganizes, augments and distributes information (RSS, blogs, wikipedia)
• Social / colloborative tagging• P2P based information retrieval• Two emblematic applications: http://del.icio.us/ and
http://flickr.com/
Differences with traditional indexation
• Form nor content of the metadata are controled
• Produced by the user community: fundamental change in the resource-user relation, where the authority of the librarian/archivist/conservator is questioned
• Incorporates metadata that are intrinsically linked to the indexer
Possibilities for the cultural sector?
• Prototype: Steve project• Advantage: « serendipity »• Desadvantage: very low semantic value of the tags• Alternative form of user-generated metadata: user
comments• Historical context • Attempt to evaluate the quality of user-generated
metadata and to draw up a typologie of these comments • Case study: image database of the National Archives of
the Netherlands
Method
• Information quality definition=> « fitness for purpose », meaning are the comments usefull to the users?
• Query analysis: compare the content of queries with the content of the comments
• Mapping with « Shatford-Panofsky » categories
Shatford-Panofsky categories:
Results:
• Categorisation of queries: S1=17,50%, S2=5,5%, S3=57%, S4=2,5%, G1=9%, G2=8,5%
• Categorisation of comments: S1=67,61%, S2=18,87%, S3=30,70%, S4=20,56%, G1=6,29%, G2=1,71%, G3=0,57%, G4=0,29%, A2=2,86%
• Queries and comments alike concentrate on specific notions, use few generic terms and no abstract terms.
Typologie of the comments:
• Corrections of the existing metadata: 34,13%
• Narrativity / context: 18,87%
• Personnal experiences: 4,29%
• Opinion: 2,86%
• Dialogue / questions: 1,15%
Correction of existing metadata
Narrativity
• Certain comments put diverse and scattered information into a context
• => Lev Manovich « Database as a cultural form » (Language of new media)
Personal experiences
• Small number of comments reflect on personal experiences regarding the image
• What is the interest to other users?
Personal opinions
• Very few personal comments
• Again: what interest to other users?
Dialog
• A small number of users poses questions and interacts with other users by sending comments
• Acts in a forum like manner• Helps in the creation of virtual communities around
heritage institutions
Postmodern indexing?
• The role of each intervenant within the information chain is no longer strictly defined: user - librarian - indexer - editor - author
• Two index-layers: the authority of the librarian/archivist/conservator is confronted with the informal and personal metadata of users
• How can these different layers be managed within a collection management system?
• « Narcissm of the viewer» should be avoided