Maybe it is time to innovate and rethink 'discovery' services rather than simply providing an open source replication of existing proprietary approaches? A major long-term study of how readers discover scholarly content noted that use of library discovery services may have peaked in 2012 and is now in *decline*. (See: 'How Readers Discover Content in Scholarly Publications'. By Tracy Gardner and Simon Inger. Renew Publishing Consultants. August 2018) http://renewpublishingconsultants.com/wp-content/uploads/2018/08/How-Readers -Discover-Content-2018-Published-180903.pdf There are other approaches. Library centric reading/resource list systems In my work with academic libraries I have noted (presented and written about) a big increase in the deployment of library centric reading/resource list systems which, at least for many undergraduates, are now the 'discovery' systems of choice. Unlike (almost?) all catalogue/discovery systems they helpfully provide a *patron context* to discovery--i.e. the course/module the student is on--even the year/week of the course. In the main (notably undergrad') students love this more straightforward and relevant approach and may never use a library 'discovery' service. In addition these library reading lists solutions integrate closely with the university's learning management system and often provide access to 'learning resources' not typically found in the library catalogue/discovery system. Titles can be annotated with additional metadata - typically added by faculty - such as 'essential' or 'background' [reading]. There is barely a university in Australia or the UK that doesn’t deploy this (complimentary) approach to discovery --though the US and other countries seem to be lagging behind. There is more information on the reading/resource list page of Higher Education Library Technology (HELibTech) https://helibtech.com/reading_resource_lists AI Yewno https://www.yewno.com/discover/ is an interesting AI based approach to providing a different kind of discovery environment. A number of libraries are using it. Yewno harvests 'millions of scholarly articles, books, and databases across virtually all academic fields' to allow users to 'navigate intuitively across concepts, relationships, and fields, learning from resources that might have otherwise been overlooked'. Voice With voice search becoming ubiquitous (Gartner predicts that by 2020 "30% of web browsing sessions" will be voice https://www.gartner.com/smarterwithgartner/gartner-predicts-a-virtual-world- of-exponential-change/ is any library doing work on this?--e.g. using the growing number of tools to optimise their website and/or catalogue/discovery service for voice search or to develop voice user interfaces (VUIs). It seems like library content and tech providers are working on this.. (AI and voice search featured at the recent ConTech conference in London) Linked data Finally there was a lot of talk a few years back about the opportunity to enhance discovery using linked data and a number of catalogues offered a linked data set - but I'm not aware of anything in place yet that looks really transformative in terms of the user experience of *discovery*. Ken Ken Chad Consulting Ltd http://www.kenchadconsulting.com Tel: +44(0)7788727845 Twitter: @kenchad | Skype: kenchadconsulting |Linkedin: www.linkedin.com/in/kenchad Researcher IDs: Orcid.org/0000-0001-5502-6898 ResearchGate: https://www.researchgate.net/profile/Ken_Chad -----Original Message----- From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of Till Kinstler Sent: 11 January 2019 15:12 To: [log in to unmask] Subject: Re: [CODE4LIB] Spit-balling - open-source discovery layer Hi, Am 10.01.19 um 23:24 schrieb Fitchett, Deborah: > To date there’ve been a number of successful open-source catalogue systems – Koha, Evergreen, the current FOLIO project. Discovery layers seem to have been left out of scope of all of these. There are several open source discovery layers for libraries, in wide use are for example VuFind (https://vufind.org/) and Blacklight (http://projectblacklight.org/). But these are user interfaces, that come with a search engine component (in these two cases the open source search engine Solr) but without bibliographic metadata. > My impression is that the main reason for this is the problem of the metadata index. Metadata is hoarded by for-profit vendors; some of them only grudgingly work with the Big Discover Layer companies under strict conditions (and possibly with money changing hands, I don’t know…) so would be unlikely to just hand it over to a community project. No metadata, no discovery layer. We are a not-for-profit library consortium in Germany and we are running a "discovery search engine" for our member libraries (with about 200 million records at the moment). This search engine (also implemented with Solr) has no user interface, bot only an API that libraries can use to connect their discovery layer to this search engine. From our experience, if you talk to matadata providers (like publishing companies, data aggregators etc.) they are usually willing to give their matadata for free, at least if libraries buy licences for their products (in general it is recommendable to negotiate free delivery of metadata for licenced material with publishers, in our experience they are willing to do that, at least for consortia or other groups of libraries). It's even in publishers' interest, because they earn money with licences and not with metadata. Putting metadata for licenced material into discovery layers is actually like advertising the stuff they earn money with. So getting the data is actually doable. But very much work goes into converting, normalizing and enriching the metadata you get from all these different sources in different formats (that tend to change over time). Don't underestimate that, that's extremely cumbersome work, at least if you want to have some "data quality" at the end of processing. > But more and more, metadata is becoming available through other sources. Crossref is the biggest and most obvious, and then there’s hundreds or thousands of institutional repositories. So my thought is, is it now becoming possible to create an index at least tolerably suitable for an open-source discovery layer, using open-access metadata sources? I know that some libraries or library consortiums are using article data from Crossref. We also looked at it and it seemed to be rather "thin". But still, far better than nothing. > And if so… how hard would it be? It's doable, but doesn't come for free. Don't underestimate the amount of work necessary to run a reliable backend for metadata provision for a dsicovery system. Till -- Till Kinstler Verbundzentrale des Gemeinsamen Bibliotheksverbundes (VZG) Platz der Göttinger Sieben 1, D 37073 Göttingen [log in to unmask], +49 (0) 551 39-31414, http://www.gbv.de/