Library Favorites and Resource Modeling

Editor's Notes

• #2: Good afternoon!Self-introduction… Web Systems Manager, oversee Drupal, Article Discovery, Proxy server.Note about NISO ODI – been taking part, many of you saw survey invitation. Looking to standardize vocabulary and processes around discovery systems so libraries, services, and vendors are all speaking the same language. Survey & report coming out in spring 2013. Talk to me later if you have questions.And now on to the main event….SENTENCEDesigning, building, and mining data from an academic library's "favorite resources" tool.FULL DESCRIPTIONWeb site visitors to the University of Michigan library can save some kinds of resources (catalog items, databases, online journals, and article citations) to their user account for future use. Users can optionally organize these resources into categories (the system recommends courses they are taking and categories they have previously used, but individuals can create any categories they like). In this session, attendees will learn about our design process (including user studies, design elements, and Drupal coding) and the usage of the tool. The pool of saved items becomes a rich data source for providing anonymized, aggregated data to library staff and site visitors. We will conclude by exploring some of the possible uses of this data, including building supplemental reading lists for specific courses.
• #4: Been at University of Michigan since 2007.Led a Drupal implementation, development of Summon article discovery with a Drupal module, personalized search interface, MTaggerCurrently serving on the NISO Open Discovery Initiative – looking to develop common vocabulary and best practices for communication between libraries, discovery providers, vendors – so we’re all talking about the same thing. Survey just closed. Report coming late winter/early spring 2013.When I submitted this proposal way back in January, the new system was to have launched in June.Well, things happen.So we launched it, but not until September 11. So usage data I’ll talk about will be smaller than I’d hoped. But it’s still interesting.
• #5: We had experience with tagging. Launched MTagger in 2008. Very little adoption. A few librarians jumped at it. “MLibrary 2.0”. Mostly, didn’t.Tagging was wrong user interactionActual needs:Save stuff for laterOrganize it logicallyTie to academic life
• #6: Mirlyn Classic still does.
• #7: Needed a quick replacement for MetalibExpediencyExplore the value
• #8: Most functionalAllowed tags
• #9: Search Tools is our brand for article discovery (ArticlesPlus/Summon), Database, and Journal Finder. Favorites is just a toggle; available in your list of favoritesOnly findable from list of Article, Database, or Journal favorites (you had to remember).
• #10: Same as for Databases and Journals.A simple list (alphabetical) of your saved items.A little bit clever; saved both a short-term reference ID plus enough metadata to rebuilt an OpenURL (for articles).
• #11: Sonali Mishra, in library’s User Experience Department, was largely responsible for this inquiry. Workgroup was colleagues Albert Bertram, Jon Rothman, Sigrid Cordell, Sonali Mishra, and myself.Used a paper prototype to validate the initial design and metaphors
• #12: Our experience; it didn’t work. We pulled tagging out of catalog and digital image collections in January 2012. Nobody noticed. We pulled it from the library web site without fanfare when we launched Favorites. Nobody noticed. Plus, nobody had tagged anything in a long, long time.
• #13: Idiosyncracies:“to read”, “ignore”, “looked at”“project 1”, “later”Random keywords of obvious significance to user, but not to the casual observer (me)We weren’t sure if the right label was “tags” or “labels”. Users had a weak preference for “tags” (1 said label, 2 said tags, 5 didn’t have an opinion). A couple said “labels” reminded them of GMail labels. Those act very differently, so we ditched that name.Liked recommended tags.
• #14: Walk through the screen, what’s where, how it works.
• #19: Two kinds of tagsMost recently used (by that user)Currently-enrolled courses (for students & faculty)
• #20: This thing creates lots of nodes: favorite items, favorite tags, and favorite (the glue)May not be sustainableArticlesPlus – no record identifier that is persistent. So we save enough data to build a citation & create a viable OpenURL.
• #22: Data generated by Albert Bertram
• #23: Data generated by Albert Bertram
• #24: Data generated by Albert Bertram
• #25: Data generated by Albert BertramArticlesPlus is our article discovery toolMirlyn is our catalogDatabases & Online Journals are from the DB & journal finder
• #26: Data generated by Albert Bertram
• #27: Data generated by Albert Bertram
• #28: Data generated by Albert Bertram
• #29: Interesting privacy concerns.Of course, data are still pretty sparseDare I say “Portal”? I didn’t think so.
• #30: Happy to answer any questions.