Friday, March 01, 2013

New dawn : the changing resource discovery landscape

Another definite case of JIBS targeting one of the most live issues for users of eresources and eresources librarians in its February 2013 Workshop, attended on Monday 25th by nearly 100 librarians, publishers and intermediaries.  The day was deliberately planned to encourage sharing of experience and ideas.  Especially so with the implementation of Resource Discovery Systems (RDS) in institutions and their experiences post go-live.  To this end the speakers were in the main from the library-side rather than the publisher/provider-side.  That said, the latter were well-represented and an excellent opportunity was provided for suppliers to hear about the impact of their discovery products as well as the sector's very real concerns.  A discussion panel at the end of the day (and at breaks through the day) allowed for some dialogue, but it was agreed much more was needed to understand how - or if? - major challenges could be resolved.  But the event was just as valuable in terms of bringing to light the very practical matters we have to take on in implementing this radically new skin on the content for users, and to shift our focus from the local to the global, which is where the keynote speaker,

Ken Chad, beginning with a Global technology trends report, took us first.  Viewing RDSs' from the perspective of global changes in technology use, trends in technology as deployed in Higher Education, and the claims of Google in the longer-term, Chad helpfully asked how or if at all RDSs were aligned to meet the needs of users in their future and changing expectations. Such a fresh look at the suppliers' offering here is important.  RDSs have arrived on the marketplace and libraries have swooped to implement, often at great speed and with good deals arranged with vendors.  They apparently provide, after all, the answer to the long-standing problem of users swarming to Google instead of to the libraries' interface on content in which institutions have made heavy investments.  Here at last is the panacea : the Google-type interface to the content libraries have selected for users.  Except, no.  The problems and the solutions are more much more complex.  By reference to Gartner's top 10 strategic technology trends for 2013, Chad unpacked what we would do well to consider for our discovery solution that went way beyond the issues we had all thought most urgent right now - not least the partiality of the RDSs' knowledgebases. Issues that might appear to be "on the periphery" of the RDSs turn out, once framed within your institutional strategy, to be core.  As Richard Rumelt would encourage us to ask in his Good strategy, bad strategy: Just how closely to the challenges you had laid down in that does the RDS match up?  The future landscape of the dominance of mobile, personal cloud computing, multiplicity of platforms, enterprise app stores, strategic big data, actionable anayltics, integrated ecosystems, quite new learning environments beg many questions of RDSs.  From the basic - is yours touch-optimized?  To the large scale -Is your RDS getting better the more it is used?  Is it enabling prediction of behaviours and optimization of service to you as managers?  Is it empowering your decision-making?  What trends are you seeing?  How have you strategically placed the RDSs' variety and complexity of data in your institutional approach to use of big data?  Where is the ownership, control and power going?  Heterogeneity and loose-coupling of systems at the start of the RDS scene is being left behind wherever opportunities arise for bringing systems back together, for vendors to have a greater control of the solution stack.  Such questions are the heart of what is actually in or out of your RDS but with the overlap between system and content getting more complex, who is driving change exactly becomes more open.  Conversely the current lack of personalization possible with most RDSs fails to keep up with innovations in the learning space - the emphasis on experimentation and play towards achieving learning goals.  Again, just how hospitable is your RDS to encouraging exploration of topics?  How imaginatively does the interface and design promote this learning behaviour either individually or in differing social groupings?  The questions seem to proliferate - to issues of integration with Massively open online courses (MOOCs) and VLEs; openness and the possibility of an emerging preference for Apple as the platform of choice and the route to discovery for etextbooks (see the Open Nottingham agenda); and the work done at Huddersfield on activity data to try to reach some conclusions about the impact of libraries on learning outcomes.

At this point Chad did a double-take and weighed up RDSs in the light of Google's future aims laid out in Wired magazine's article on Google's future. This is where we could really have done with some discussion of the RDSs' relevancy rankings.  Not so much how they work precisely - which is going to remain hidden for as long as we're tapping on the Search button - but how they stand up against Google's ambitions for its Knowledge Graphs and sense-making of search through use of artificial intelligence and natural language understanding.  But the room at SOAS remained eerily quiet after Chad had raced through the implications of Google's aims and his review of current activities in HE around discovery (Discovery.ac.uk; KB+; RISE; Meaning based search at MIMAS; Huddersfield's Lemon Tree).  Chad suggested "Things not Strings" was the take-home message of Google's remaining the pioneer, the elsewhere the user will prefer to go if RDSs don't shake up and offer a real alternative in terms of both content uncovered and in experience of how successfully the user's problem (rather than "search" per se) is understood and interpreted.  But it was his sentence about search - "What once seemed transactional now seems an extension of ourselves" - was most striking.  And most telling about the differences to be overcome still in the space libraries offer their users to search in.

We then heard case studies on implementation of RDSs from different institutions.  Emma Crowley from the University of Bournemouth where their implementation of the Ebsco Discovery Service (EDS) was branded "MySearch", spoke of how it had "changed my life".  Her and Bournemouth's journey was so familiar to many - the sad fortunes of Webfeat, with users sceptical about its promise to revolutionize the search experiece almost before it had had a chance.  Difficult, slow, "particularly useless" for many academics, unpromoted, disappointing, geared to markets in the US and Australia, the best that could be said about it finally that it was "Only useful to who which other databases you should search!"  So by the time SerialsSolutions had taken it on was too late for Bournemouth.  Another solution was needed, and fast.  A good and well established relationship with Ebsco assisted in the choice of EDS and the familiarity of users with the Ebsco platform.  There was an easy integration with Bournemouth's existing products (Talis catalogue; Eprints repository, Blackboard VLE) and Ebsco was keen Bournemouth should contribute to EDS's development, becoming a Beta partner with the University of Liverpool.  Popular and so important for Bournemouth was integration of their BU CHAT widget, which they were also able to work with Ebsco on, using a template API Ebsco had developed.  

In their implementation, Bournemouth took some decisions that stemmed from a desire to introduce the EDS rapidly and not to disrupt users, keeping branding used from Webfeat (MySearch) to minimize confusion, display a single view to benefit interdisciplinary searchin, and against forcing prescriptive subject searching.  Over their previous offering, users like best how super-fast MySearch is.  They appreciate how you can add more resources to those already searched; link through to full text; set up alerts; displays results immediately while continuing to search other resources; get the BU CHAT widget (for chat with a librarian on queries); and link to referencing software quicker than normally.  There are concerns with academic colleagues who want greater exposure and clarity from the system about what is searched, and indeed the institution wants students to understand the breadth of content they have searched - this is expected in assignments and moreover shows the student return on the investment they have made in their education.  The technical challenges Emma described appear to be lightweight in the main, with good help from Ebsco.  The issues here were more with authentication and the very regrettable fact vendors have not kept pace with the transitioning from Athens MD to Athens Open.  More serious are the data issues and inaccuracies in matching to licenced content.  Data in the knowledgebase reflected US licencing for Books 24/7 for instance, and US licences for journal bundles rather than Nesli2.  Emma concluded the one-stop shop the library could now provide did make a real impact for students and for librarians the sheer dullness of inducting students in multiple database searching was a welcome thing of the past.  BU had moved from clunkiness to usage stats so impressive for EDS they spoke for themselves.  "We get better value from our subscriptions - and see a signficant increase in sourcing and use of metadata from databases that previously had little use".  You can't argue with the usage stats.  But some interesting questions emerge: Should BU default to full text only?  Students expect full text so should BU show them that first?  This makes sense, but students will then be limiting themselves away from the full breadth and depth of data available.  Also, should BU integrate all it can offer?  Rather than, as presently, selecting from the data sources available?  Will that just lead to showing items of questionable relevance making for a Google-type  information overload?  Has BU got the balance just right already, avoiding disappointment by thinking carefully about what it includes?  The metadata itself though can skew results nastily (e.g. in the EDS's display of results from Scopus).  The ongoing inconsistencies in 3rd party platforms create barriers for userslinking out to full text and there gaps in coverage – law and market research - that grieviously impede full take-up by those schools.

At the University of Cambridge, Isla Kuhn, from the Medical Library, presented for Emma Coonan who runs the Research Skills Programme, and herself. In a "game of 2 halves", we learned how a pilot of the Summon system from SerialsSolutions/ProQuest at Cambridge had affected the training of library users in research skills in the humanities, arts and social sciences and in the hard sciences.  The library world can be a really alien place, and for Cambridge students one that can be overwhelmingly complex and large.  An arcane catalogue that imposes restrictions left right and centre and categorizations upon things the whole time can be the last straw when you've been given a reference mid-lecture that is crucial you do read, but which remains only that thrown on the air mid-sentence ("Read Smith, 2000").  Thus shed-loads of time is spent explaining how to cope with this Martian construction that is the catalogue - why can't I search it for journal articles? - which can now be spent on what is genuinely helpful, evaluating resources and thinking creatively, and with pleasure rather than pain, about searching and the assignment.  It's far from a binary system that would split users between disciplines and make that relative to the usefulness of the RDS; nonetheless, in any subject field that requires any rigour in search - absolute reliability in recall and precision plus ability to return specialized data types - its going to fall down and badly.  This was illustrated powerfully in a contrast of a newspaper article Back pain - just ignore it, and the research article on the Wiley Online Library that inspired it.  In Medicine, it is fundamental that everything is found, a systematic review is done, and Summon is just not suitable for such rigor of interrogation.  In the waves of popularity for RDSs, the rave reviews and sales-talk, you might need to issue a self-apology for wishing to preserve the Boolean search, but it is undeniably the case that the individual databases (covered, partially covered, or not covered at all in the RDS) retain their place when the academic discipline requires it.  This is certainly true for the postgraduate user, but equally so for the undergraduate in those disciplines.  And it's not just about retrieval of stuff per se and in quantity.  A large part of the exercise of literature searching is to establish how new your research question really is which means quantifying exactly where and what the gap in the literature means.  Another image that delighted the audience (earlier Isla had referenced the It's a bit more complicated than that T-shirt and Dave Patten's "Users should not have to become mini-librarians") was a football pitch showing Summon used as part of a team - with multiple other datasets and search tools in position on the same pitch. 

Paul Stainthorpe then gave a really entertaining, lively and fascinating account of the practicalities of implementing the EDS RDS at the University of Lincoln, branded Findit@Lincoln.  The unexpected effects of dropping discovery into the system pond leads for everyone to post-implementation problems you could never predict and that represent major pieces in the project, but which are largely unseen and silent.  Which is why it was so helpful and revealing to hear about them.  Paul faced getting close and personal with systems, standards and data that he had not had to encounter before working for 10 years with library systems. Dealing with a typically aging set of systems not talking to each other, his environment was typical - maintaining too many confusing routes to content, too many barriers to full text, too many users complaining about not being able to access eresources, a federated search tool singularly failing to address the problem.  The worst of it was that users though being good "mini-librarians" accepting their sacrifice, doing obediently what subject librarians told them to do, still couldn’t find their full text.  With a renewed impetus that saw the launch of a new Website, new authentication system and a new system to support reading lists, Lincoln implemented the RDS with an upfront search box on the library home page that users couldn't miss.  Meanwhile, backstage, unexpectedly, systems and standards not heard of for many years as that relevant in the eresources space, from Z39.50 to (dare it be said) MARC21, were picked over to understand why links were not functioning properly and the exposure of library data was causing such headaches. 
Though presented as in some way niggling or minor, the issues Paul enumerated had significant knock-on affects for users, from the display of dates in differing conventions (month-day; day-month), to locations and collections that no longer existed, broken URLs in records for ebooks, failure to employ the all-important facets in the system if the data that drove them was lacking or faulty.  All in all, Lincoln had been "positively disrupted" and the importance of clean data for use and re-use across multiple systems was now thoroughly understood; and crises in cruddiness of data had led to improvements in workflows such as the move to a file export of MARC records that rather than involving a complete rewrite of all catalogue records, instead detected new and changed records only and incrementally wrote them to the export.  The solving of issues locally in institutions must be going on separately but RDS vendors should look to KB+ and seek to adopt the best practices for authoritative data enshrined in that project.  Other challenges Lincoln has faced include the use of data from the intstitional repository which is not uniformly full text; the absence of signficant databases covered by the EDS (which for subject librarians is the first question asked), and how to manage the very activity of problem-solving when new problems are getting ever more complex and difficult to understand.  For many the idea that you "don't remember anything" because you're fire-fighting all the time was completely familiar and true.  As at Huddersfield, Lincoln is interested in the use of activity data to uncover what how it can inform the links between library service and degree results; and looking to the future, the better integration of the EDS with reading lists and the EDS's inclusion of personalization tools for the user.

Elizabeth McHugh re-focused the meeting after lunch on issues of cost, training and liaising with the RDS supplier and their support mechanisms.  It is always important to involve the right teams, bringing cataloguers in and ensuring good buy-in from your parent institution.  Futureproofing in the context of the UK looking hard at shared services is key, ensuring that the RDS chosen can accommodate multi institutional subscriptions in consortia and changes that may be made in and outside such consortia.

Adam Edwards continued with an overview of the implementation of Summon@Middlesex.  Picking up on what we had earlier, Middlesex found time became available for focusing on the skills of evaluating resources  to promote deep learning now search had been made so much easier and faster.  And the refining tools and choices for full text online and scholarly publications online were really popular.  In an immortal phrase for the ease of use for generating references Adam quoted one student: "That is so well sick man, innit!".  However, sadly there is a big However: Not everything is indexed that Middlesex subscribes and the idea of moving to a universal solution is still some way off.  Law (e.g. no Westlaw) and Business studies (e.g. no Keynote) suffer particularly (and Health for which the Cochrane reports are not covered).  This is really serious as people vote with their feet and an entire school just opts not to use Summon@Middlesex, period.  Forcefully it was related how Ebsco will not release the data to Summon and so Ebsco's resources are not covered.  This leads to a situation of worry and doubt from various parties that undermines the product, becoming a corrosive problem that ultimately leads students right back to - guess where - Google.  Middlesex, and they are not alone, resort to tactics such as demoting Ebsco resources to the end of a list so students go to the other resources first.  And the alternative of subscribing to ABI Inform (ProQuest) has been discussed as alternative to Business Source Complete (Ebsco).  Quite the wrong drivers then determine the decision-making on resource selection: "we shouldn't have to have conversations like this".  Yet another alternative - from desperation to desperation - is to have two RDSs - EDS and Summon.  Analogies from the business world often throw things into relief in libraries so you can see problems for what they really are.  And Adam Edwards didn't stint from this approach.  He stressed the importance of word of mouth recommendation and the fluidity of constitutencies of interest - already the Dean will prefer Google Scholar.  Try and sell hard a system that is this partial that "doesn't quite work".  And this in the context of student fees where institutions need to demonstrate the tangible benefit from their investment. 

Fiona Greig from Plymouth University talked about her institution's diverse student body, its size and spread across the world and how the multiple systems and designs put in place to date could get in the way of students making use of materials.  So Plymouth's implementation of Primo and Alma was as much about improving the experience for the user as it was to address the need to drive down back office costs. What Plymouth found was immediate acceptance by new students of the new service but some academics were
"very unhappy" and wanted students to use specific resources (apparently even if they were included in Primo).  Some staff had problems adjusting to the importance of filtering searching and refining.  And again, the partiality of subject coverage meant some schools were disadvantaged.  But at least it was clear federated search had been buried and was now dead and the path was set to improve Primo.  Greig's presentation was instructive especially in terms of the dual implementation of Alma and the benefits that had had upon performance of Primo.  The outsourcing to the cloud meant immediacy of update to all holdings and availability information for instance and production of analytics on data use that would only encourage more and deeper integration of managing differing types of resources this way.  Like Adam Edwards, Greig agreed the question of who is now in control is a live one, having major repercussions on discovery.  Less prosperous publishers need to be engaged for their data or we could see the disappearance of resources from existence simply through failure to index them ("If it is not easy to find why pay for it?").

Finally John Dalling from the University of Wales Trinity Saint David spoke about the implementation of OCLC WorldCat Local where the solution of the RDS also addressed harmonizing  two sites (at Lampeter and Carmarthen), two library management systems and two catalogue interfaces.  Like Primo and Alma at Plymouth, Worldcat Local is hosted in the cloud, no server is required on campus, and no big IT support is needed.  The process to harmonize with the WorldCat catalogue though was a brave step since if no match of the local database with WorldCat is found no record is shown (this affects about 2% of the records).  The benefit though is enrichment to the local record from the WorldCat record.  Similarly the move to working with OCLC Connexion's client has meant both benefits and compromises.  Usage data for instance can now be extracted easily and clearly shows increase in use - for instance of JSTOR and the ebooks packages - as a result of the implementation, but sharing the knowledgebase in the OCLC model means searches returning multiple editions of a book will prioritize the edition held in most OCLC libraries which invariably is not the edition held in Trinity St Davids.  While the harmonization of the different instances of the data across the two campuses has been improved, the batch processing of records has to integrated with WorldCat Connexion and to date this is involving much manual work because of the input of staff needed to do Connexion database maintenance.

Matt Borg from Sheffield Hallam ended the day with a discussion of how implementation of Summon there had brought "cultural changes" in the student experience and in the library - student interface and relationship.  Echoes were heard with Chad's opening - the basic expectation so often previously frustrated that things "make sense" to the user.  The "old discovery" started with a view of the federated search or the library catalogue.  How could we have ever thought these were good enough?  Or any good.  The OPAC seemed to come out of a Hewlett Packard factory - designed by engineers for engineers - a tool for expert users to use.  Summon, Borg didn't mind admitting, took a massive step away from all that but still - as social media showed - got users mad: "Library gateway is pissing me off tonight".  How close have we got to Shoshin in our RDS?  The beauty of the beginner's mind that should really influence its design "the Mind of the Beginner is Empty, Free of the Habits of the Expert, ready to Accept, to Doubt and Open to all the Possibilities".  The usability testing at Sheffield Hallam focused, then, on expert listening rather than any other expertness.  Using screencapture software it could be verified that students did indeed find stuff they never did before : "I never found stuff before you put the Google box on the library gateway".  For information literacy training, it has meant the "biggest change ever". The arcane complexities training in hundreds of databases involves are gone for good.  Thank goodness.  This leads almost to a sense of deskilling since if the interface doesn't need explaining what is the point of induction?  Instead there is the opportunity for making the experience enjoyable so that learning points stick.  And this is a really big cultural shift to be celebrated.  Now you can focus on your behaviour as a searcher - are you a Panda or a Magpie? - and in such awareness do you want to morph to be an Eagle or a Lizard?  Anyone interested in this aspect, the move away from the mechanics of search, should check out the blog on Information Literacy & Summon