17

Taxonomies 2 0 2008 craig rees v1-1

Embed Size (px)

DESCRIPTION

A presentation on the difference between social tagging and controlled vocabularies for information rich businesses

Citation preview

Page 1: Taxonomies 2 0 2008 craig rees v1-1
Page 2: Taxonomies 2 0 2008 craig rees v1-1

•  Why do we care about Metadata?

•  Where do folksonomies and controlled vocabularies come into the equation?

•  Who uses what?

•  What do I think the silver bullet is?

•  Questions

Page 3: Taxonomies 2 0 2008 craig rees v1-1

3

Australia’s leading information resource Helping you find, buy and sell

•  12.5 million consumers each month •  600,000 advertisers

Page 4: Taxonomies 2 0 2008 craig rees v1-1

•  Commercial Manager for Content & Search –  How can we improve the search experience –  How can we make the most of our content

•  Involved in the industry since 2000

•  In the past –  Ran product management at BBC new media –  Advised UK media and ecommerce companies on content

management and search strategies

Page 5: Taxonomies 2 0 2008 craig rees v1-1

•  Controlled vocabularies have concepts / terms

•  Folksonomies have tags

•  Metadata is the association of either terms or tags with a piece of content

Page 6: Taxonomies 2 0 2008 craig rees v1-1

“Metadata: cataloging by those paid better than librarians “ Rot Tennant, Points of Pain, Peculiar Possibilities, & a Patron Paradise 2003

Page 7: Taxonomies 2 0 2008 craig rees v1-1

For content centric organisations

•  Search

•  Content

Page 8: Taxonomies 2 0 2008 craig rees v1-1

Folksonomies Controlled Vocab

  Low management overhead

  Requires a user population who want to contribute

  Requires a significant volume of data to filter out noise

  Relatively low initial investment

  Poor experience on day 1

  Significant management overhead

  Needs a team of experts both subject matter and information architecture

  Can be costly in comparison

  Requires significant investment in infrastructure and management tools

  Rich experience on day 1

Wisdom of Crowds Wisdom of Authors

vs

Page 9: Taxonomies 2 0 2008 craig rees v1-1

•  Yellow™ –  2,865 headings, 150,000 terms

•  BBC –  Over 100,000 terms

•  The bigger the taxonomy, the harder it is to find the correct term

•  Sophisticated management systems are required

•  Need specialised skills

Page 10: Taxonomies 2 0 2008 craig rees v1-1

•  Keeping the CVs up to date and relevant

•  Language differences even within the same country

Page 11: Taxonomies 2 0 2008 craig rees v1-1

•  Business intelligence –  Search analysis –  Inferred user folksonomies

•  Advertiser input

•  Market analysis –  Industry trends –  Trade associations –  Trade publications –  Local government –  Legal requirements

Page 12: Taxonomies 2 0 2008 craig rees v1-1

Web service based automated tagging solutions

•  Open Calais www.opencalais.com

•  Tagthe.net tagthe.net

•  Inform www.inform.com

•  Gracenotes www.gracenote.com

Page 13: Taxonomies 2 0 2008 craig rees v1-1

•  Spelling mistakes, incorrect use of terms (my perspective or yours)

•  Gaming and user manipulation

•  High volumes should mean that wheat is separated from the chaff

•  But this takes time…

•  And what do you do in the interim period

Page 14: Taxonomies 2 0 2008 craig rees v1-1

Folksonomies Controlled Vocabulary

•  User generated

•  Decentralised controlled

•  Publishers •  Centralised

control

Page 15: Taxonomies 2 0 2008 craig rees v1-1

•  It’s about combining the wisdom of authors with the wisdom of crowds

•  Languages change, new terms evolve folksonomies and user led terminology is at the front of the curve

•  Use both, sometimes in different circumstances –  Front end search improvement vs back end content aggregation

•  Create a feedback loop to aid self improvement

Page 16: Taxonomies 2 0 2008 craig rees v1-1

Review

Controlled vocabulary

Listing

Folksonomy

Authors associate metadata at the point of content

entry

Content is searched via associated metadata

Users tag content as appropriate

with their own terms

Content is aggregated

based on the metadata

associated

New terms are reviewed and where appropriate gaps in the controlled vocabularies are filled

tagging and search reports are run to track

usage

1 2

3

4 5

6

External information sources

Search results