Aubrey Madler, Rural Asssistance Center
40% of search failures have to do with vocabulary failures
Types of controlled vocabularies:
- Synonym ring
Synonym ring: dog, canine, mutt, pooch
Hierarchy: UF, BT, NT (used for, broader term, narrower term)
Thesaurus: Medical Subject Headings (MeSH) from NLM, for example
Creating a thesaurus
1. Generate wordstock – subject experts, users, publications, your org.
Top-down: start with existing thesauri
Bottom-up: start with users
2. Decide on format, choose preferred terms and identify synonyms. Consistent about capitalization. Usually use plurals. One term per concept.
3. Choose hierarchies and facets. Facets are ways to refine search. In shopping site, for example, color, brand, price range.
4. Add associative terms: related term, see also. Both terms should be in the thesaurus, but you want to help the user find both.
5. Select thesaurus design and display. Electronic or print? Staff only or visible to public? Facet, hierarchy, or both? Topical order or alphabetical? Visual or strictly text?
It’s never finished. New terms come up all the time. You’ll notice gaps, which could show in search logs. The collection changes. Society, literature, fashion change. Are you going to reindex the collection?
Planning and maintenance. Document what and why you did.
Melissa Rosales, TBWA/Chiat Day, and Andrew Carlos, The Harker School
Folksonomies: quick and dirty
Allow anyone to add tag. An impromptu bridge. It’s a cost-effective way of dealing with a large amount of information.
Trust power user who is more knowledgable.
“Ugly tags”: people don’t know the “rules of the house.”
Tags identify who it’s about, what it is, who owns it, refining categories, qualities and characteristics, self-reference, task organization (e.g., toread, towatch).
Serendipity: “I found it!”
Tools like Google Reader, Google Alerts: have to preload them.
Browsing tags is somewhat similar to shelf-browsing. Tag browsing, though, is more proactive.
The Wikipedia Game: find connections from one thing to another.
Social bookmarking: Delicious and others. (Zootool, Pinterest, The Fancy)
Del. more visual than it used to be. Zootool allows sorting by media.
Social media more than just text and sharing, heavy interest in visual media.
Issues and problems:
- Polysemy (multiple meanings)
- Basic level variation
Users like to tag in the singular, because they are thinking of the one item they are tagging.
Some users are very granular, specific (“young adult paranormal romance”)
Moderator: what about combination of thesaurus and folksonomy?
Melissa: One case found 92% original content from users.
Aubrey: LC did that with their photo collection.
Audience: RLG study on user-generated content. www.oclc.org/research/activities/aggregating/
Aubrey: Home-grown content management system.
Andrew: Has tags enabled in his system (catalog?)
Audience: Uses medical subject headings, but borrows from LC, nursing terminology. Has trouble tagging personal items. Some days in a more general mood, some days more specific.
Moderator: LC subject terms don’t always match subject matter experts (curators at his museums)
School librarian: Students take notes. Tagging makes more sense after you’ve thought about the subject, looked at subheadings, etc.