considering new discovery layers tom adam, information literacy librarian, university of western...
TRANSCRIPT
Considering New Discovery Layers
Tom Adam, Information Literacy Librarian, University of Western Ontario
Alan Bell, University of Waterloo
Nora Gaskin, McMaster University
Sian Miekle, University of Toronto
Martha Whitehead, Associate University Librarian, Queen’s University Library
Tom AdamUniversity of Western Ontario
What we’re doing…
• Why?
• Variables and options
• ILS and the Discovery Layer
• Interface Decisions
• Are we there yet?
What’s it all about?
• Embrace change
• Work backwards• Consider the user
What we know about what we do
How we are perceived…
Our Brand…
What would you do?
the User Experience…
• OCLC
• Perceptions survey
• Extract issued 2006
Where do you start?
89%
I use a library website.
IT’S RIGHT HERE
& IT’S 2%
I use a search engine like google.
COST-EFFECTIVEEASY TO USECONVENIENTFAST
0
10
20
30
40
50
60
70
80
90
CREDIBLEACCURATELibraries are… Search Engines are…
Reasons …
CR
ED
IBL
E
AC
CU
RA
TE
RE
LIA
BL
E
CO
ST
-EF
FE
CT
IVE
EA
SY
TO
US
E
CO
NV
EN
IEN
T
FA
ST
RELIABLE
Search engines fit perfectly with
my lifestyle.
Our Legacy…
What can we do?
Make the search engine better…Canadian Undergrad
Martha WhiteheadQueen’s University Library
Discovery Layer? Next Generation Catalogue?
Discovery Layer
ILS - circ
Images
Theses
Vendor Vendor VendorILS – bib, holdings
E-J
Fed Search
About “discovery”
• From inventory to discovery (and back again)
• Keyword searching versus index browsing
• Relevance ranking: precision, recall and first page principle
Open Source or Vendor
• open sourcee.g. VuFind, eXtensible Catalog, SOPAC
• ILS vendorse.g. Primo (Ex Libris), Encore (III)
• other vendorse.g. Endeca, BiblioCommons
AquaBrowser, OCLC, Serials Solutions
Local or Hosted, Customize or Not
• Local: – hardware/software infrastructure– some local configuration possible
• Hosted: – minimal local hardware/software – ideally a service to manage user generated
content
• Either: data matters
Web 2.0
• Why?– interaction with collections– reliance on peer and expert opinion – social search
• What? Where? How?– user generated content
• academic concerns, placement, web scale
– usage logs, recommender systems
Summary: some factors to consider
User experience degree of integrationsearch intelligenceextensibility
Local (or consortial) capacity time and money
Sian MiekleUniversity of Toronto
Introducing the Discovery Layer
• Data loading
• User experience
• Local vs.hosted solutions
Data loading decisionsShould data be loaded in new system?
Advantages:Self-contained interface Data available to functionality of new system
Challenges:Shifting the dataSynchronizing the data
Dis
cove
r L
ayer
Ap
pli
cati
on
ILS
Build indexes
8.5M items 6.4M bib records 2.3M authorities 144K MARC holdings
Restart application
Convert to XMLfor search & display
Data transfer (nightly)
XML• <RECORD>• […]• <PROP NAME="p_title">• <PVAL>The Canadian almanac & directory.</PVAL>• </PROP>• <PROP NAME="p_publication">• <PVAL>Copp. Clark [etc.], 1848-</PVAL>• </PROP> • <PROP NAME="310.a">• <PVAL>Annual.</PVAL>• </PROP>• <PROP NAME="362.a">• <PVAL>[1st]- year; 1848-</PVAL>• </PROP>• <PROP NAME="503.a">• <PVAL>Title varies: 1848-50, Scobie & Balfour's Canadian almanac, and
[…]</PVAL>• </PROP>• […]• </RECORD>
Not quite 1000 MARC fields, 26 subfields, but…
“Convert to XML”
MARC data
<RECORD> […] <PROP NAME="p_title"> <PVAL>The Canadian almanac &
directory.</PVAL> </PROP> <PROP NAME="p_publication"> <PVAL>Copp. Clark [etc.], 1848-
</PVAL> </PROP> <PROP NAME="503.a"> <PVAL>Title varies: 1848-50,
Scobie & Balfour's Canadian almanac, and […]</PVAL>
</PROP> […]</RECORD>
01/30/09
Search strategy overview Jan 22-28 2009
New search
50%Facet refinement
search
42%
Search with no results
4%Search within previous
results
4%
User experience: using facets
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
% of 1st facet % of 2nd facet % of 3rd facet
Relative importance of facets Subject geographicalareaSubject time period
Call number range
Genre
Author
Subject
Publication year
Language
Format
Library
User experience:changing search model
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Author Subject Call number
Structuring the search
Postsearch
Presearch
User experience: feedback
• 169 comments, Sept 08 – Jan 09• 2:1 in favour of the new interface• 50 comments requesting features (26 ideas)
– 6 ideas based old search models 6 grads, 3 library staff
– 6 ideas for new features4 grads, 2 undergrads, 1 library staff
– 14 ideas: functionality missed from old system 6 faculty, 12 grads, 12 library staff, 3 undergrads
Local vs. hosted solutions• Time to think• Flexibility
– Content: BIP, web pages, other data repositories
– Presentation: visual, multiple, changing
• Challenges– Infrastructure– Staff
• Shared local solution?– Share common tasks– Share resources
Nora GaskinMcMaster University
1. User interface choices & decisions
2. User feedback & behaviour
UI Choices: Basic Search Screen
• just a search box & go?
• include choice of search indexes? (fielded searching)
• include alpha as well as keyword indexes?
• include limiting at time of search?
Search ‘n’ Go: UWO (Encore)
Search ‘n’ Go: Queens N.Y. (Aquabrowser)
Search ‘n’ Go & Limit by Collection: Vanderbilt (Primo)
Choice of Keyword Indexes: U of T (Endeca)
Choice of Keyword Indexes & Databases:
Villanova U. (VuFind)
Choice of Keyword & Alpha Indexes: Florida State U. (Endeca)
Choice of Keyword & Alpha Indexes: McMaster (Endeca)
Choice of Keyword/Phrase Indexes & Limit by Format, Location:
Waterloo (Primo)
UI Choices & Decisions
• Search, search within results, start over – Oh my!
Search Within Results; Search button; Start Over link
New Search button; Search Within Results button
UI Choices: defaults
• facets: open, closed or a mix?
• after a closed facet has been opened, keep it open?
• after a search, keep the previously selected index chosen, or return to the default?
UI Choices: facets
• which? hierarchical or flat?
• what to call them?
• display order
• order of values within facets: alpha or most hits at the top?
• all “and” or some “or”?
UI Choices & Decisions
• Search logic– match on any 1, 2, 3 terms, all terms?
• Display– how to display the full bib & item details
• Alpha indexes– how to display?
1. User interface decisions2. User feedback & behaviour
McMaster’s Initial Implementation
• Spring 2007
• usability testing
• feedback via “Tell us what you think” button in catalogue
• most feedback within first 3-4 months
Some interesting resultsfrom usability tests...
• only 1 of 5 subjects used the dimensions (oh no!)
• some subjects relied on type-ahead to complete all queries
User Feedback: Negative
Suggestions/Problems/Questions
User Feedback: Positive
Most-Used Facets, Jan.-Dec. 2008
Alan BellUniversity of Waterloo
The hope, the promise
• One search box to provide fast and convenient access to all of our resources, regardless of format– Enable access to our digitized cultural memory– Address disjointed library experience– Attempt to remain relevant to web savvy users– Provide Google-like relevance and react to challenges
on the commercial internet
Considerations, challenges
• Existing OPAC features • Becoming an aggregator
– Can we ever get metadata/access to everything?– Native interfaces– A lot of data
• Controlled vocabulary cross walks • Local modifications and needs• Partners/Community and Collaboration
http://animaledventures.com/2007.04.01_arch.html