IWMW 2009: Hands-on prototyping for (meta)data structures
Download as PPT, PDF0 likes505 views
UKOLN supports expertise in digital information management. This document discusses prototyping techniques for metadata structures, including free-listing terms, consistency checking, and card sorting. Card sorting involves having users group terms into logical categories to understand how people conceptualize an information structure. The results can be analyzed to evaluate how frequently categories are used and inform the design of classification systems and conceptual models. Prototyping techniques like card sorting provide a user-centered perspective to test and improve metadata structures at an early stage.
IWMW 2009: Hands-on prototyping for (meta)data structures
- 1. A centre of expertise in digital information management www.ukoln.ac.uk UKOLN is supported by: IWMW 2009 Hands-on prototyping for (meta)data structures Emma Tonkin Talat Chaudhri Alexey Strelnikov
- 2. Introduction: metadata What is it, why, and who cares? Actually covers quite a lot of things: Terminologies (taxonomies, controlled vocabularies) Vocabularies Data structures/models (entity- relationship models, conceptual models)
- 3. Example For example: − 'Event Metadata' excerpt Event Type: Workshop Workshop Location Workshop Type Workshop Date (ISO 8601/W3CDTF) Opinion of Workshop {Boring, Interesting...}
- 4. Eliciting information from users People have many different opinions, and are capable of believing a lot of mutually contradictory ideas! :-) Classification systems or data structures (hopefully) represent a consensus or at least a compromise They therefore improve as we learn more about opinion They also exist (and are useful) within a given context of use – for a given purpose
- 5. Organisation... Needs context! And exists for a purpose! Clean sock drawers are very nice, but don't necessarily achieve anything unless you wear socks :)
- 6. Prototyping techniques Scope what we're trying to achieve! Learning about data structures Building quick easy models Finding easy ways to test them
- 7. Free-listing 'Name all the x's you know' − eg. What types of image can you think of? − Or - What types of book are there? − Or - What features might you use to describe a book? − Or – What sort of resources do researchers create and publish? − You might have your own use cases to look at!
- 8. Free-listing on “workshop metadata” Peer review status Date Times Duration Event in which it occurs Venue Attendee list Major trends, topics Organiser(s) Remote attendees? Streamed? Outputs? Feedback Presentations Papers Slides Tutorials Related articles, links, people Coffee Breaks Networking (who met who) Networking (wifi!!) Pitches Endorsements (SIG, etc) Other events on similar topic Other events run by same people... Research area and findings Related research activities
- 9. Consistency Build a table: − (Note: This is a lot like social tagging!) Compare and contrast your results. Word Appearances Venue 5 Related research areas 2 Breaktime beverage 1
- 10. Card sorting Once you have some terms, then it's time to figure out how they fit together This is especially easy to do with taxonomies − supports interface design by grouping functions/menus/etc, − Supports development of conceptual models
- 11. Card Sorting: Method 1) Make sure you have something to sort (using for example the approach described earlier) 2) Bribe/trap some 'users' into agreeing to take part (note 'user' means 'the sort of person you are trying to write a system for') 3) After explaining the rules ('we're not testing you... just learning about the area') ask them to sort the cards into logical groupings May be 'open' (users can create novel groupings or categories) or 'closed' (users must use predefined groupings)
- 12. Using the results! Another table can be built from these results: − Which categories were created? − How frequently were they used (how many users applied them?)Term/ Category Bird Nature Manmade Food Buildings Seagull | || Lighthouse | | Fish & Chips ||
- 13. Evaluating an information structure (eg classification/taxonomy) Scenario-based evaluation: 1) Get existing classification 2) Develop some scenarios for which it can be used 3) See if it is possible to step through the data structure in order to successfully complete a scenario Example: Card sorting
- 14. Final comments A user level view of things is probably simplified (and so it should be!) What the user sees does not have to be what you store... What the developer sees probably does – good documentation helps to encourage uptake of data structures/metadata. Test early, test often. Card is cheaper than code!