Access 2006 notes: October 12

My continuing summaries from Access 2006. Thursday, October 12th was the first "normal" day of the conference featuring the following presentations:

• Open access, open source, content deals: who pays? (Leslie Weir)
• Our Ontario: Yours to Recover (Art Rhyno, Walter Lewis)
• Improving the Catalogue Interface using Endeca (Tito Sierra)
• Lightning talks
• Panel: Towards a Canadian Digital Information Strategy (Susan Haigh, Lucie Molgat, Brian Bell, Bill Maes)
• Progress and Prospects for European Digital Libraries (TEL – European initiative)

` <>`__

` <>`__Open access, open source, content deals: who pays?

` <>`__Leslie Weir, VP/President-Elect of CARL

Open access mandates from research funders such as CERN, and near mandates from research funders such as SSHRC, are encouraging

` <>`__University of Ottawa:

• disappointing experience with BioMed Central which touts itself as an open access journal (following the "author pays" model)
• UofO had subsidized their researchers until this year, when BioMed demanded a 15x increase in their funding
• but researchers are now campaigning for a return of the subsidies

` <>`__Open source modules as an alternative to the traditional commercial ILS

• Libraries have been early adopters of technology, starting with mainframes, through PCs and Internet access
• ILS reflects the need for integrated information
• While our OPACs were originally cutting-edge, they've been relatively unchanged since their inception compared to rich catalogues offered by commercial retailers
• Ex Libris touts Primo as a way for libraries to unlock the potential of their information
• McMaster announced on September 6th plans to adapt the same search and browse technologies underlying Home Depot's catalog system (Endeca) to provide faceted search and browse

` <>`__Role of Scholar's Portal in OCUL vs. individual OPACs:

• Should we implement a shared ILS?
• If so, should it be a vendor's ILS, a collaborative open source ILS, a new layer of discovery tools over top of a traditional ILS?

` <>`__Content deals for consortial arrangements:

• Canadian libraries invest heavily in consortial access to e-resources
• OCUL invests approximately $20 million annually for 20 university members
• Problem is how to divide the costs:
• OCUL in 2002/2003 tried to adopt a five-year plan for a "user-pay" model based on the amount of use of each product
• Now in the fourth year of the plan, many institutions were experiencing 20% increases (capped maximum) year over year that was not sustainable, while others were experiencing significant decreases
• Budgetary certainty has quickly become a stronger requirement for the formula, while continuing to refine the balance of ability to pay vs. use
• CKRN began with a grant for $59 million for traditional 'hard' sciences
• On November 4th CKRN is going before an international panel to propose a $48 million project to move into social sciences and humanities

` <>`__How to deal with the administrator who questions the value of library services in the age of Google?

• OCLC has determined that the vast majority of the information that is available on the Internet is not freely available
• Libraries have a role in providing the keys to that information
• Accepting that not all roads lead through the library and integrating with other services like Google Scholar will help lead users to the information they need
• Google has raised the bar; it's no longer okay to tell our users "These are the 22 interfaces you need to learn to find the information you need"

` <>`__Our Ontario: Yours to Recover

` <>`__Art Rhyno, Walter Lewis

` <>`__Knowledge Ontario

• Knowledge Ontario grew out of the Ontario Digital Library and is now consituted by six projects; Our Ontario (search.ourontario.ca) reflects community content
• Hope to use Lucene as the "digital hammer" to search community content (pulling from newspaper records, historical society data, etc)
• Excellent at merging indexes, so potential for libraries to use their regular workstations to index in parallel
• Excellent separation between indexing/searching layer and presentation layer
• OurOntario includes digitized books (searchable via additional metadata), fonds (indexing EAD), images from Images Canada (stripped-down DC), TEI documents
• AlouetteCanada is a layer on top of the same Lucene backend using SOLR to present facets

` <>`__Great Lakes Images

• Great Lakes Images presents pictures with Google Map integration to show where the pictures were taken
• Includes images donated from individuals, questions raised by KnowledgeOntario staff, and comments from users that actually correct or enhance the metadata for the records
• Museums etc. often don't make the high-res version of the image available for free, and that's fine with KnowledgeOntario

` <>`__Local stories

• Full community newspapers that are ignored by most aggregators or services
• Essex Free Press includes PDF directly from the production system, with fully searchable text that is extracted from the PDF

` <>`__Improving the Catalogue Interface using Endeca

` <>`__Tito Sierra, NCSU Libraries

• NC State was the first library to use Endeca as a front end to the library system
• Endeca is a software company in Cambridge, MA that is a software provider for Home Depot, Indigo, and others
• McMaster has just announced that they will be using Endeca
• Project details: http://www.lib.ncsu.edu/endeca

` <>`__Motivation:

• Improve the user experience when using the library catalogue keeping the backend system the same
• Exploit the existing authority infrastructure possible within the MARC record format

` <>`__Why Endeca?

• Relevant result ranking (previous system provided "most recently added to the catalog" approach)
• Faceted browsing narrows search results, or starting with a browse:
• Facets include language, LC classification, subject genre, subject access, author, most popular (based on circulation data)
• You can deselect a facet you chose previously at any given time
• Performance and speed
• Search "comforts":
• Spell correction (auto-correct)
• "Did you mean..." (prompts user)
• Stemming
• Sort options (publication date is most popular)

` <>`__Relevance ranking controlled by order of ranking modules:

• Original query match
• Phrase match
• Field match (tiered -- multiple submodules)
• Number of fields matched
• Weighted frequency
• Publication date descending
• Circulation date descending

` <>`__Interface was built from scratch on top of the Endeca search software and indexing:

• Benefit is that NCSU has local control over the UI to enhance it based on user feedback
• URL is persistent -- so you can copy and paste the URL resulting from a search and share it with others

` <>`__Features not supported:

• Work level aggregations / roll-up (no ability to show the item rather than the expression)
• Customization
• Folksonomies
• Recommendations

` <>`__Technical overview:

• Co-exists with SirsiDynix Unicorn ILS and Web2 online catalog
• Still uses Web2 authority search
• Export MARC records
• Endeca parses the MARC records on a nightly basis
• Resources: 5 IT staff, 1 cataloging librarian, 1 reference librarian
• Timeline: License in spring 2005, production in summer 2006

` <>`__Implementation challenges:

• Some feedback suggests too many facets have been exposed
• Relevance ranking is hard
• Faceted navigation across logical groupings (like item type of audio-visual for CDs, tapes, etc)

` <>`__Usage statistics:

• Most popular facets:
• Subject topic
• LC classification
• Format
• Keyword search (default) is most popular by far

` <>`__Usability test compared 5 users on old system with 5 on new system

• New catalog was proven to be easier and faster on 9 out of 10 tasks

` <>`__Reflections

• Post-launch, new interface required backend data cleanup as many interesting variations were exposed by the facets
• Search and discovery were "broken" with the old interface

` <>`__Lightning talks:

` <>`__John Derno, UVic

` <>`__Building a backup catalogue in a week and a half

• Facing a four-day Voyager upgrade & outage
• Considered WorldCat, but it only allows 16 simultaneous users
• Zebra indexed 1million records in 40 minutes on a reasonable development box (1GB RAM)
• Basic PHP interface through PHP/Yaz to talk to Zebra via Z39.50
• Major challenge was documentation of indexdata.dk technology, e.g. building queries

` <>`__Godmar Back and Annette Bailey, Virginia Tech

` <>`__MAJAX or "Look Ma, no server"

• OpenURL services provide a link to check catalog for searches that don't work
• Desire to integrate live OPAC information (MARC records, holdings records) into web pages in your organization's domain
• JavaScript library provides an AJAX interface to III Millenium catalog
• Does not require any changes to Millenium installation

` <>`__Jeremy Frumkin, OSU

` <>`__LibraryFind

• attempt to build a federated search tool that is not slow
• the approach: cache search results
• version 1 built with PHP & MySQL, will be thrown away with version 2 (RoR)
• layout deliberately designed on Google interface, user testing shows that the adwords column gets ignored (no surprise there)
• does describe related terms
• linked titles are available in full text, link for AJAX-y link resolver that no users clicked
• show/hide affordance also wasn't used
• Jeremy and Mr. T (including full photoshopping of Jeremy with a Mr. T haircut)

` <>`__David Fiander, UWO

` <>`__How I spent my summer as an Evergreen developer

• Georgia Public Library consortium decided that it would be cheaper to devote 6 people to developing their own ILS rather than continue to pay a vendor
• System did slow down the day it launched due to the immense load
• Development team works entirely online (email and chatrooms), with sporadic meetings
• David wrote the code that enables the self-checkout terminals as a separately available open source module
• Koha would like to offer self-checkout, David will probably adapt the module for them

` <>`__Terry Reese, OSU

` <>`__Intro to MarcEdit

• Written in C#, works well on Windows, reasonably well under Mono on Linux and Mac
• Converts between MARC and various XML formats
• Converts between MARC-8 and Unicode
• Includes a Z39.50 client
• Designed to be programmable rather than forcing you to use the GUI
• Possible cataloging tool replacement for SmartPort / WorkFlows?

` <>`__Marc Jordan, SFU

` <>`__"Shameless self-promotion"

• Mark wrote a book Putting content online: a practical guide for libraries, Chandos Publishing
• Wanted to make it fully open access, in the belief that making it available would encourage sales of more copies
• Two sample chapters are online

` <>`__Walter Lewis, Windsor

` <>`__What do we do with content that's too big to fit on the screen?

• Frequently occurs with maps, high-res photos
• First option is to provide a small version of the photo, but that doesn't give proper treatment to the real image
• Digitization engine behind is Mr. Sid (proprietary, LizardTech) -- gives a Google Maps-like ability to drag around
• Mr. Sid is geographically aware, so it knows which way is North on a map
• You can index X-Y offsets that would make it possible to deliver historical map pointering
• JPEG2000 standard includes some x-y axis capabilities

` <>`__Towards a Canadian Digital Information Strategy

` <>`__Susan Haigh, LAC

More information -- http://collectionscanada.ca/cdis

` <>`__Why the need for a national strategy?

• current efforts are fragmented, uncoordinated, under-resourced
• incapable of preserving what we create
• inconsistent across the provinces
• weak in comparison to other countries

` <>`__Why now?

• enough accrued experience
• volume of digital content is rising sharply
• methodology is becoming clearer, storage costs are coming down

` <>`__Who or what can it serve?

• research, elearning, community building, e-democracy, public accountability

` <>`__Canada's Scientific Infostructure (Csi): Universal, seamless and permanent access to information for Canadian research and innovation

` <>`__Lucie Molgat, Director Csi Program, CISTI

` <>`__What is CISTI?

• Canada's national science library
• Has experience providing seamless access for NRC users across Canada
• Publisher of 15 journals

` <>`__Csi vision

• National information and infrastructure network to provide seamless and permanent access to full-text digital content

` <>`__Csi components:

• Infrastructure, repository, and services and tools
• Using an SOA to enable partner projects
• Csi's repository currently has 5.7 million articles licensed from providers, will have 7 million by spring 2007

` <>`__AlouetteCanada

` <>`__Brian Bell, AlouetteCanada

• http://alouettecanada.ca
• Single-line summary: provide views over top of a comprehensive collection of regional and local digital repositories
• Addendum: variety of methods to encourage more content, including making a Web interface available for small regional organizations, Digimobiles for sending out trainers of the trainers and technology to sites that need assistance, and harvesting / normalization of existing repositories

` <>`__CIDL

` <>`__Bill Maes, CIDL -- http://www.collectionscanada.ca/cidl/

• "We are bereft of content"
• YouTube didn't need ten years of vision statements and meetings about infrastructure to get up and run
• CIDL protocol from 10 years ago is fairly applicable today, but needs to include more than just libraries
• One of the major problems so far has been a lack of content
• Need more products like OurRoots with actual content
• CIDL encourages AlouetteCanada to be inclusive -- to get everybody in the game -- but don't wait for another year to go by; DO IT NOW.
• While CARL has been an important part of startup support, AlouetteCanada needs to distance itself from CARL to encourage more partners to join in the effort
• But another goal is to get Canadians to actually view this content, and making access to the content is only part of the problem; getting people interested at even a fraction of the rate that Canadians have been downloading YouTube videos is critical.

` <>`__Progress and Prospects for European Digital Libraries (TEL – European initiative)

Ron Davies

` <>`__MACS (mapping subject headings across languages)

• English (LC), French (Rameau), and German (SWD)
• Currently have 30,000 mappings in total, 3,000 mappings across all three languages

` <>`__BS 8723 revises UK/ISO standards for thesauri

• Includes guidance on mapping between different structured vocabularies and ontologies
• Includes advice on standardizing electronic exchange