Folksonomies: Cooperative Classification and Communication through Metadata

Folksonomies

Background: Metadata
- Professional Cataloging (MARC, Dewey, LC, LCSH)
- Author created (DC, XML)
- User created: citation analysis, PageRank, weblogs, rec systems, Amazon review
Explicit User Created Metadata
- Del.icio.us: social bookmark web service
- Users freely "tag" web sites, uncontrolled vocabulary
- Flickr: photo sharing, adopts free-form tagging
- Vocabulary of users (rather than authors, professionals) reflected in the system. Dubbed "folksonomy"
- Like-tagged items collocated in the system (linux)
- Most popular tags listed, related tags on sides (automated)
Folks + Taxonomy = "Folksonomy"
- More like "categorization" than "classification" (Jacob 2004)

Popular Tags (Delicious)
- subject descriptors like: software, design, programming, music, politics, web, news, blog, css, linux, art, osx, java, mac, blogs, reference, fun, python, games, tech, photography, humor, tools, delicious, rss, firefox, toread, comics
- form/genre: comics, humor, fun, photography
- other: toread, wishlist
- occasionally authors, author nicknames
Popular Tags (Flickr)
- photo subjects: cat, friends, dog sky, sea, park, kids, garden, baby, building, flower, flowers signs, sculpture, city, vacation
- 25% of top 150 were place names
- colors, years
- cameraphone, moblog, fotolog
- "cute" and "me"

Limitations
- Ambiguity ("filtering" tag, acronyms)
- spaces, multiple words
- Synonym control and the lack thereof
Strengths
- Browsing vs Searching, serendipity
- Desire lines, quick vocabulary changes, user vocabulary
  - Merholz article, "folksonomy" vs "ethnoclassification"
Why This Works at All
- Low/No barrier to Entry, cognitive costs
- Feedback - tight feedback loop (see results of tagging immediately)
  - individual incentives - self organization ("me")
  - group / community incentives - sharing ("cute"), expert status, define terms, viewpoint, communication
Unanticipated Uses: sometaithurts, flicktion, iraq
Further Research / Conclusion
- Quantitative Tag analysis (power laws) / qualitative user analysis
- Applicability to other systems
- Best and worst of information organization: utterly chaotic, uncontrolled, but supremely responsive to user needs and vocabulary