- TagsSessionReport

Tags Session Report - 6 October 2006

Original session page is at http://dublincore.org/accessibilitywiki/TagsMetadataSession

see below for references and links to presentations

Douglas Campbell's Notes

[remote presentations]

From spectator to annotator – Seth van Hooland, ULB

Image description is very time consuming/expensive
Can improve access points with low-cost user-generated metadata such as folksonomies, also user comments
But insufficient research – analysed Netherlands image database of 500,000 images
Evaluation – fit for purpose, using Shatford-Panofski categorisation
Most queries by geographic, also important is particular people/groups/objects. Very few of abstract terms
Comments mostly on specific (people/groups/objects), some general, very few abstract
Comment types:

34% correcting metadata: 40% spelling, ??% who, 25% what, 16% where, 9% when
19% narrative to help put into context
? include personal experiences – need research to determine if of long-term value
Very few personal comments – these are used a lot in Amazon but users may not feel have as much relevance in cultural archive
Few Dialogue – pose questions and interact with other users in the comments

Need more research but can have positive impact on the metadata quality
Discussion

Any malicious comments? Seldom, most wanted to improve and share their experience

Tagging in art museums, Jennifer Trant, steve.museum consortium, USA

Monet’s Water Lillies in Metropolitan is well described, but does not include the word “impressionism”!
Is tagging scholarly documentation. Need to consider accessibility and engagement that museums are undertaking anyway – considering multiple points of view
Steve is an experiment for tools and to research tagging and resulting accessibility – steve tagger open source
Different people see different things in same work of art
Trialling presenting in groups likeness, theme, challenging
For one work: 268 terms from 138 unique words, 130 found relevant, only 10 out of 130 were in the record, ie. significant 90% were new! words people think when look at that work are not usable for discovery
So documentation we provide may not be resulting in the accessibility we are aiming for due to this semantic gap
Research project over next two years looking into this further. Incl. comparison of tags in multiple environments against dictionaries, records, image databases, and the work itself (are they appropriate in the mind of professionals)
Do they indicate teachable moments, does it involve/reward visitors more
Expect tags will always co-exist, not replace, nor reduce description efforts
http://steve.museum
Discussion

Similar issue in education – gap in terminology between teachers and professional learning theory/technologists
Are tags online different to if onsite? [lost connection to speaker]
Often find more put online results in more physical visits
Agree comments are useful, but question of how encourage users to actually comment? Sometimes have to work to prevent them!
Some let you add FOAF/vCard so get idea of the kinds of people visiting/commenting
Most common tag was “black dress”, is that really be using that as an access point? Are people doing something conceptually different when they tag. On Flickr people do to help future retrieval, but this appears to be “comment on this painting”. But sometimes the words are used to re-retreive something they remember seeing before, so these phrases are what they recall about them (or know they can use to recall them), so may be more personally related than general

Wild Metadata, Johnathan O’Donnell

David Hawking – research on university website (with mandatory metadata) found subject metadata doesn’t help people find things. Often worse than title, and only slightly better than searching the text of the URL, can its creation be justified? Searching anchor text in hyperlinks improves hits from 10 to 50%
Metadata in the wild – blog entries, tags, anchor text (easy way to collect brief description in foreign language, and know pointing to your page)
Collecting it – del.icio.us, Google and Yahoo allow you to search for links to any page or site, but unfortunately don’t show the anchor text, blogdigger provides metadata in RSS
What use it for? Quick way to find topic metadata, reality check, use for teaching how users think; auto – use XSLT to embed the metadata into the page so tells you how people are cataloguing the current page (Conal Tuohy, NZ)

Discussion
Maybe topic metadata isn’t bad, it’s as you become more professional you become “dumber” at assigning topics
Professionals aren’t necessarily doing just for public – one target audience is for other professionals/researchers, so is appropriate for them
Need to recognise difference in domains, eg. museum, gallery, libraries (may see briefly

Stu’s takeaways

Not only increase accessibility but also get engagement (which is ongoing concern)
Expect co-existence, not replacement
Professionals dumber – 90% terms not used by professionals

Terminology Services, Eric Childress, USA

MS Office 2003 has sidebar available for IE lookup whilst in Office
Looking at providing a taxonomy for users to help with tagging. Unsure if will be used, but might be some demand for geographic terms?

DCMI Bibliography, Corey Harper, University of Oregon, USA

Has 700 terms, adding 150/year, output in multiple ordering
Would like the important papers to bubble up to the top, not by official choice but by users
Perhaps rating, tags, etc
People asked for list of readings relevant as prerequisite to conference – authors could eg. tag “prereadingDC2007” so they’d be easy to find
Discussion

Could use del.icio.us? or Connotea?
Look how Pete Johnston uses del.icio.us – combines DC term with tag (eg. so know if is about or by that person)

Note: agreed to use 'dc2006' as tag for DC-2006 materials in Flickr, del.icio.us, etc.

Links to presentations

Note that these files contain the full presentation so they are quite big. We are investigating both a list for further discussion and archiving of the files....