Transcript Document
Summon Holy Grail or leaky cup? The Holy Grail • Improve the research experience • Simplicity (or, hidden complexity) • • • • • • Complete integration of print and electronic Easy to learn ( + instruct!) Easy access to article-level resources Easy to authenticate Remove barriers between libraries & users Compete more effectively with Google • Promote the use of quality-assured information • “Returning the researcher to the library” (Library Journal webcasts) • • Meet expectations / stay relevant Return on investment An unholy mess? • • • • Library Catalogue Federated search Vertical search A&I, FT Summon – what is it? • Web-scale discovery (unified discovery) • • • • • • • • • Google-like (speed, convenience, practices & presentation) Sounds like WorldCat but… Looks like vertical search but… Single search box access to full breadth of collection Centralised index of pre-harvested content Built with entirely new technology Based on extensive end-user studies Developed with libraries and content providers Promises to be a great leap forward • Simple idea • Google model • What took so long? Summon – who’s involved? • Serials Solutions • Head start (content & functionality): • Discovery & management services • KnowledgeWorks, 360, Ulrich’s, RefWorks, COS, AquaBrowser, WebFeat • ProQuest • • • • CSA, UMI, Chadwyck-Healey, Dialog, SIRS, eLibrary Experience hosting systems Relationships with content providers Development team & process • Team – Microsoft, IEEE, VuFind, Google, MediaLab • Agile development • Speed & quality • • • • began development Autumn 2008 - launched worldwide beta testing in January Liverpool: baby steps to beta in 8 weeks > open beta within 8 weeks > live service? Responsiveness to feedback Weekly updates Summon – how does it work? Summon – how does it work? • The Unified Search Index • Pre-indexes everything • • • • • • • local data taken into Summon index methods: FTP, OAI, USB, etc. formats: XML (EAD), HTML, PDF, delimited ASCII, MARC, etc. importance of good data Data normalised to Summon schema Relevance ranking of books and articles together Search architecture • Apache Solr • • • • • • • • Powerful extension of Lucene Advanced full-text search capabilities Optimised for high volume web traffic Standards based open interfaces Scalability – other search servers Flexible and adaptable Extensible plugin architecture Support for dynamic faceted browsing and filtering Summon – how does it work? • Ingesting local content • e-Resources entitlements • Client centre / SFX holdings export • Bibliographic data • FTP > MARCXML > mapped to Summon schema • local Summon server - daily update feed • check item status – API • • • • Archival data – M2M services (SRU, Z39.50, OAI-PMH) Institutional Repository – OAI-PMH Interface – Ajax, css Authentication – in front / behind? • • barrier to searching…90% drop-off in use …but Terms & Conditions • A&I providers • Summon Summon – how does it work? • “The mega-index of content” • • • half a billion records 6,000+ publishers 50,000 + journal titles • • http://www.serialssolutions.com/assets/publications/Summon-represented-titles.pdf scores of scholarly publishers and university presses • launched in January with ProQuest, Gale, Springer, Taylor & Francis and SAGE • LexisNexis, Publishing Technology (IngentaConnect), ThomsonReuters (Web of Science), ABC-CLIO, IEEE, Emerald, Scitation, The Royal Society • ...and growing by the week • MLA International Bibliography, Ingram Digital (MyiLibrary), PsycArticles and PsycCritiques, M.E. Sharpe, Knovel, IBIS, RMIT, Hart, Allen & Unwin • Create a movement of publishers and aggregators • Critical mass • Discovery or obscurity? Quick & simple Quick & simple 0 – 4.3 million in 2 seconds! Advanced search Results display Results display - articles Results display - books Item status check - incl. multiple copies Results display - archives Refining results • Full-text filter • Scholarship filter • “Mega-index” Refining results Extent of collections at a glance Refining results – more options Refining results – more facets Saving & exporting Saving & exporting Saving & exporting Saving & exporting Other features • Times cited • • • • “Did you mean?” suggestions • • British vs. American spelling differences Mobile app (for iPhone, Blackberry, Android, Palm or Windows Mobile) • • • Other significant resources Web of Science – others? usage-based recommendations / bX Recommender? Auto download Full-text Open API • • pre-selected search targets in departmental web pages course management software modules In summation • Feedback • • • • • Usage stats? Summon & informal mechanisms University – unanimous approval Library staff – curbing enthusiasm Performance • • • Speed & simplicity - vast improvement Consistent results, clearly displayed Relevant results • Book rankings / newspaper noise • Simple, obvious starting point • Still value native interfaces...for how long? • • Content, tailored indexes Summon has the superior search engine – more thorough? • De-duplication • Powerful results management In summation • Coverage • Breadth and depth but... • Effectively represent academic libraries in UK? • US site: • • • 99% of top 100 downloaded titles 100% of top 50 JSTOR titles 100% of their top 50 titles from OCLC databases • Law, Science Direct (Ingenta metadata?), EBSCO • Much more useful than federated search • Content is critical • ...or what happens about unharvestable content • Will federated search ever go away? • Coverage • Greater transparency of what’s being searched • how do I know what’s there, from where? In summation • Access to full-text • Summon / SFX interoperability • Cannot use date to resolve (but ArticleLinker does) • Newspaper articles • SFX cannot link into CSA • • STM workaround - Scopus and Pubmed Arts and Social Sciences? • Two-character first names (eg. Chinese name Li) • • SFX inserts full stops > searches for L.I. > search fails! Dead ends - dissertations and patents • Neither SFX nor ArticleLinker seems to work – where next? • Access to native interfaces • • Subject access still required? Summon ‘Recommender’? • which resources the most / best results come from • link to native interfaces In summation • Alternatives - none available but... • EBSCO Discovery Service (by end of 2009?) • • • • • • • • Unified + federated search Scaled-up version of their existing search engine Demo? / facets? “7 times as many 3rd party content partners as any other discovery service” subscription agent - established relationships with publishers aggregated full-text database products are larger than ProQuest’s less embargoed content Ex Libris – Primo Central • Unified + federated search • Content providers? • Deep Web Technologies (fed & unified) • Battle for supremacy http://liv.summon.serialssolutions.com