6 "inside-out" activities librarians are doing
I expect that most readers of this blog are familar with Lorcan Dempsey's idea of distinguishing between "outside-in" vs "inside-out" activities for libraries.
I have argued that with open access gaining ground, the traditional role of academic librarians where they own or acquire access to books and journal articles for their community (the outside-in approach) is likely to decline in importance. Instead librarians are increasingly focusing on "inside-out" approaches.
http://blog.impactstory.org/oa-by-when/
"In the inside-out model, by contrast, the university, and the library, supports resources which may be unique to an institution, and the audience is both local and external.... Often, the goal is to share these materials with potential users outside the institution."
It has been just under a decade since this concept was first mooted, a interesting question to consider is compared to say 2010, has the % of the library workforce in academic librarians shifted significantly to inside-out from outside-in ? What % of your FTE (full time equalvants) are now devoted to such tasks today as opposed to 2010? How would the balance shift in 10 years? 15 years? In fact, there has even been suggestions to add this has a new metric to monitor!
(As an aside I'm pondering over whether all librarian activities can be covered under just these two types of activities)
But before we even do this, I guess one thing to do is to list some of the inside-out activities that academic librarians do today.
Here are some common and not so common roles that academic librarians play today that probably fall into the inside-out role.
Also see "Inside-out library services" by Mark Dahl that goes into detail on areas like Digital Scholarship which I'm less familiar (hat-tip Lorcan Demsley)
1. Managing institutional repositories (IR) and data repositories
In a way librarians have always helped to publicise faculty output even before the digital era. Many libraries like mine may do book displays or exhibitions of books authored by our researchers, though this only allows limited exposure locally.
With the internet, we can now do the same thing on a global scale with institutional repositories.
This seems to be the role that has caught on the most in academic libraries. While the success of institutional repositories in getting their researchers to deposit their articles can be disputed, today the majority of academic libraries host their own institutional repositories, from cloud hosted solutions like Digital Commons to locally hosted open source platforms like Dspace.
Many if not most of such institutional repositories also host their institutions' electronic thesis and disertations collection , which is one of the greatest yet often less mentioned success stories of such repositories.
It was not too long ago, that access to electronic thesis was often only available via expensive DDS/ILL service or by subscribing to Proquest Dissertations and Thesis. Today, if you want to access a recent Masters or Phd thesis , there is a high chance you can find it by googling and finding it in a institutional repository (barring rare cases where it was embargoed by the author).
With regards to data repositories, things have not progressed so quickly, but I see this year as a turning point, where adoption of data repositories by institututions appears to have turned a corner but hopefully academic libraries have learnt from the past and not adopt a "build it and they will come" mindset.
Returning to institutional repositories, I think the history of IRs has shown that building repositories, sitting back and expecting researchers to deposit their output particularly journal articles is an approach doomed to failure even if backed with institutional mandates that are toothless. As such , a lot of energy is now expanded by Scholarly Communication librarians on the promotion and marketing of the service to researchers and this I think constitutes a big part of the inside-out activities of modern academic libraries.
As a recent Open Access Button presentation at CNI Spring notes, it is important "to build the simplest way for researchers to deposit an article and a best in class system for outreach". This includes publishing guides like Direct2AMM which provide easy to follow instructions for authors to obtain an Author Accepted Manuscript fromjournal submission systems, figuring out best ways to contact authors to get them to self archive , tools to quickly check ILL/DDS requests and more
Open Access Button presentation at CNI18S
The problem of discovery
Another point I think that is less appreciated until recently is that once you have obtained the output, it is not as simple as putting it in your repository and expecting discovery to happen automatically.
George Macgregor Institutional Repository Coordinator at the University of Strathclyde has done quite a lot of work, trying to optimise discovery of his IR using various SEO techniques in Google and Google Scholar
It is telling that he notes in his literature review that "it is perhaps surprising to note that only a limited amount of prior work has been documented on repository discoverability approaches and their evaluation"
I'm not surprised Macgregor did not find much, my personal impression is that enhancing the discovery of IRs has been relatively overshadowed by the issue of trying to fill up repositories with content.
Indeed even as recent as 2012, there was an article documenting the difficulties some institutional repositories had with exposing all their content in Google Scholar, though since then the Google Scholar team seems to have engaged with IR contributors/developers of major systems, and this has greatly reduced these issues (assuming you are not stuck on early versions).
Indeed things are looking much brighter today, there are projects like The Repository Analytics & Metrics Portal (RAMP) in the US , IRUS-UK that focus on measuring the impact of IRs.
While Google and Google Scholar are still dominant when it comes to discovery, the success story of Unpaywall and similar services (dubbed by some as access brokers), has lead some libraries to start optimising for other sources.
For example University of Liège Library released "Open Access discovery: ULiège experience with aggregators and discovery tools providers." They consider sources like BASE, CORE, Primo, Summon , Unpaywall and more.
The same issues of discovery are going to start emerging around datasets as well, hence the importance of understanding data citations, how articles link to datasets (Scholix) and following the impact of Google Dataset Search - the Google Scholar for dataset.
Google Dataset Search Webinar - everything you always wanted to know about Google Dataset Search
2. Collecting , curating and preseving special collections (including digital Humantities Projects)
Academic libraries have began to pay closer attention to their own unique collections, to curate, digitize and preserve them.
My own institution for instance, creates and creates Oral history recordings around the pioneers who were heavily involved in the founding of our young university.
The National University of Singapore Libraries has started to curate Digital Humantities projects by their researchers, While the National library board (Singapore) digitize maps collections for spatial discovery
I see also a role for libraries in encouraging and curation of OER (Open educational resources) .
All in all, this is a new emerging and challenging role. OCLC research calls this trend of going beyond the collection of books and journal articles to collecting more inputs, processs and outputs as "the evolving scholarly record"
Remember the issues I mention around discovery of Green OA articles? Those difficulties pale in comparison to handling Digital Humantities projects, OER and other less traditional content types, which we have little or no experience in handling.
3. Bibliometrics work, RDM and reproducibility
Today the modern academic librarian is pretty at home with bibliometrics and often assists the institutional research office as well as researchers in obtaining and analysing research metrics such as Citation counts, FWCI, H-index & more.
I never thought this work fits into the inside-out view until I realised that University & individual researcher rankings today are heavily influenced by research metrics and librarians by advising the community (including advise on research profiles and metrics) can have a impact on them .
Case moved up 86 spots in the Leiden Ranking by doing bibliometric clean-up in Scopus etc. One librarian had this impact! No one else on campus has ever moved the institution that much that fast! #cni18f
— Lisa Hinchliffe (@lisalibrarian) December 11, 2018
Today the concept of Open Science, Research Data Management and even reproducibility is increasingly becoming an area of concern. Many academic libraries are taking a role in these areas, and it is common to see Research Data Management Librarians appointed and in rare cases even Reproducibility librarians.
By helping researchers prepare their work, in particular data sets for sharing to the community outside the institution, we again see the inside out role emerge.
One of the more interesting questions of course is whether libraries in general are supporting such initatives library wide, or are they still considered the purview of just a few Scholarly communication librarians that make up a vanishing small percentage of the workforce?
University of Eastern Finland Library for example "set up a research support and open science services project in the library, with almost 30 people (about half of our total staff) participating."
4. Helping optimize researcher faculty profiles and presence
We live in the age of researcher profile proferation. Depending on how you define one these include Google Scholar Profile, ORCID, Researchgate profiles, Academia.edu, Mendeley profile, SSRN profile, CRIS profiles, Vivo, researcherID (soon to be depreciated) , Scopus AuthorID and more.
Some academic librarians hold workshops on how to navigate this confusing jungle of systems and more importantly how to manage their online presence.
Others focus on systems they manage and control such as CRIS systems (eg PURE, Converis, Elements) , or setting up ORCID (institution might be institutional member).
However let me sidetrack a bit and talk about a pretty new idea I came across in this area.
Setting up research profiles in Wikidata
Recently I came across an idea to setup researcher profiles in a pretty novel system - Wikidata.
Talk by librarian at Wikicite 2018 on IUPUI's pilot on creating researcher profiles in Wikidata
In a fascinating talk at Wikicite 2018, the Digital Intitatives Metdata librarian at Indiana University-Purdue University Indianapolis (IUPUI) , talked about an experimented to create 19 faculty profiles for a small department - the IU Lilly Family School of Philanthropy in Wikidata.
But what is Wikidata? It is essentially a project to generate structured link data for use in Wikipedia.
The idea here is to express statements in "triples" for use in Wikipedia. Triples if you are not familar express statements in the form "subject-predicate-object" or informally it a way to state that "Something" has relationship "Something else".
So for instance Wikidata will have statements that says
Aaron Tay is an instance of a human
Or
Aaron Tay has affiliation Singapore Management University
(Q50359431) (P108) (Q950609)
As Wikipedia has many language editions, each of the elements above are assigned a alphanumeric code. So for example Q50359431 represents me, Q950609 represents Singapore Management University and P108 represents the employer affiliation.
You can of course add a lot more statements, for me. For example you can use properties like gender (p21), educated at (P69) and even identifers like ORCID (P496), Scopus Author ID (P1153) and much much more
In fact there is something special about humans, you can use triples to create statements on literally anything on the sun.
But since we are talking about scholarly profiles, we are interested in creating statements about articles and books and you can do that too.
Here for example is a Wikidata item with statements about my first published journal article - Improving Wikipedia's accuracy: Is edit age a solution? ((Q58004730)
You can see that this Wikidata item has statements saying
a) It is an instance of a journal article
b) has the title "Improving Wikipedia's accuracy: Is edit age a solution?"
c) has the subject "Wikipedia"
Mostly importantly it has the property "Author" of Q50359431 (mouse over my name) which is me! v
This is the essence of linked data, creating relationships between items. It is unfortunate this article was imported without references otherwise you might be able to see the references it links to using the "Cite" property P2860
To learn more about Wikidata and Wikicite (the group that focuses on biliographic material in Wikidata), I recommend the following recent Wikidata 101 talk at Wikicite 2018.
How much Scholarly data is in Wikidata
You might be thinking this sounds too much work, who is going to add in all these data. In fact, this very article I showed? It wasn't created by me (though I did edit the author property to link to my entry), the data was already inside Wikidata!
As of now (Dec 2018) there are 20 million publications (including articles) in Wikidata already and this is 40% of total entities in wikidata! You can also see we have over 160m citations, thanks to the open citations available on Crossref and other sources.
Another stunning fact about journal titles in Wikidata.
There are now 42,650 items about scholarly journals in @wikidata, up +80% since last year.https://t.co/a14c0ssbc7
— WikiCite (@Wikicite) October 20, 2017
Still I'm not going to understate the amount of effort the IUPUI librarian did just to create the profiles. You can read about the work done here . The upshot it seems to me is that the tools that exist are still pretty raw, while there are tools like SourceMD to batch load (often helped with tools like openrefine) missing papers by PMID/DOI/PMCID or leverage ORCID to add papers to authors, it can be a slow process.
For instance to get the most out of visualizations (see later), the IUPUI librarian had to create coauthors of papers and citations in a mostly manual way though tools like author resolver tool or author disambugition tool . If you are interested here's a workshop on bulk uploading of data (by adding ORCIDs) using a combination of Openrefine and ORCID apis.
Why create profiles in Wikidata?
As you can see doing such work is not trivial and the obvious question is why borther to do this?
The first reason is that what you are creating is linked data, and it is said that Google's Knowledge Graph draws from Wikidata. Still I have not yet seen any definite evidence this helps improve visibility.
Still linked data has value on it's own and using SPARQL queries (a search method to query linked data) you can generate answers to questions that you can't using normal citation tools.
Dario Taraborelli, Director, Head of Research at the Wikimedia Foundation gave a great overview on open citations (record talk) in the recent Workshop on Open Citations 2018 and more specifically in pme part of the talk, he talks about how excited he is about the potential to answer questions that are very hard to answer today using traditional citation tools.
Presentation - Remixing the graph
Here's a simple example, can you generate citation counts of the top cited woman in Denmark? Heck if the data exist you could specify search criteria based on how properties such as educated at institution A and or any other property that exists.
SPARQL queries may look scary but in actual fact, by modifying the many examples available, it's not as hard as it looks. But even if you don't want to learn SPARQL queries many point and click visualization tools exist that work on linked data - in particular Scholia is designed for just this purpose for academic related wikidata items.
For particularly complete profiles you can see also sorts of visualizations. Some are pretty standard like papers with highest cites but you can see interesting ones you could never see in conventional tools
For example using the doctorial advisor (P184) property, Scholia is able to generate "academic trees" showing Academic genealogies, tracing the impact of intellectual traditions.
Lastly, there seems to be indications that catalogs are beginning to enhance their record displays with external linked data sources. In particular Wikidata is starting to become a very popular source with library linked data sources.
Wikidata was the data source with the biggest increase in usage among 104 #linkeddata projects and services from 23 countries surveyed by @OCLC Research over a 3-year period (2015-2018).
“2018 International Linked Data Survey for Implementers” https://t.co/s8JG0bG9Fe @KarenS_Y— WikiResearch (@WikiResearch) December 5, 2018
The IUPUI paper talks also about a project by UW-Madison Libraries - the Bibcard project that enhances catalog records by pulling in other linked data sources including Wikidata.
With Exlibris making noises about supported linked data in Primo, could we see Wikidata further increase in importance?
Resources from Wikicite 2018 :
Wikidata 101 & data modeling - Slides / Recorded talk
Ingesting data in Wikidata - Slides & Handout / Recorded talk
Querying Wikidata - Slides
In particular if you are looking to set up Scholarly profiles in bulk for your institution, you should definitely look at this handout that shows you how to use OpenRefine 3.0 to extract ORCIDs for your institution using the ORCID API, reconcile with wkidata entries and export back into Wikidata to enhance or create Wikidata items with properties like Affiliation as well as ORCID . Once you have that, you can feed a list of these DOIS to SourceMD, which will bulk query Crossref to create wikidata entries for articles belonging to those ORCIDs and link to the author wikidata items. Then use the new author disambiguator tool to do further clean up (remove short names) and add coauthors.
5. Exposing collections (including catalog records) with linked data (eg Wikidata)
Besides creating Scholarly profiles in Wikidata, one can also leverage wikidata for special collections.
Organizations such as the Biodiversity Library and the National Library of Wales have all started to expose and link their metadata to Wikidata. In particular, "Having overseen the curation of detailed Wikidata about the book trade in Wales, including authors, publications, publishers and distributors, the Library now has an ambitious plan to share metadata for every book ever published in Wales, of Welsh interest - a plan which could see the creation of around half a million Wikidata items. "
I would add that other players have also talked about their work with Wikidata including National Library of Sweden, National Library of France, OCLC etc. In particular some like OCLC, National Library of France are adopting Wikibase the software behind Wikidata for production cataloguing workflows.
OCLC Linked data Wikibase prototype
About the project by OCLC, they "partnered with 16 academic, research, public, and national libraries to prototype the reconciliation service – to connect legacy bibliographic information to linked data entities; and an editor service – to view, create, and edit linked data descriptions and relationships. "
You don't have to be a national library or big academic library in the future to expose your data to the web via schema.org or convert your catalog from MARC to Bibframe
Services like Library.Link Network that provide collaborative Linked Data publishing services are available.
http://library.link/#relevance
Given, this is the first time I have heard of this, it's surprising how fast it has grown.
Talk - Connecting Wikidata to the Library.Link Network at Wikicite 2018
6. Library as an open citations provider
This is probably stretching but I have recently started to notice a few libraries attempt to go beyond merely digitizing maps and to start extracting citations from the Scholarly record and releasing it in a open semantic citation format.
The first project that caught my eye is the Linked Open Citation Database (LOC-DB) in Germany.
You can read the paper or watch the talk below but it is a very interesting project by a few German Libraries to scan print books and process electronic journals in areas such as social sciences (in German) , extract the citations (which requires state of art citation extract algos for extraction) & process them (eg reconciliation , validating and linking to items in Crossref etc) into “standardized, structured format that is fit for sharing and reuse in Linked Open Data contexts”.
You can see a similar initative in Switzerland, where there is an attempt to build a collaborative citation index for the Arts & Humantities. It is envisioned to have two parts a "Scholar Library" and a "Scholar index". The first is just a digital library (sadly not open access due to copyright issues) but the later is a interface for browsing of extracted citations and open access.
Similar to LOC-DB , this is envisoned as a collaborative effort (currently there are 5 libraries involved).
A prototype the History of Venice - Venice Scholar Index is available.
This project is very similar to LOC-DB, though a slight difference is this focuses on extraction of references from primary sources & the availability of an interface to browse and view extracted citations.
It's a open question on whether extraction of citations is a scalable or sustainable by libraries, and this is one of the open questions the LOC-DB project is exploring.
The upshot is at the current rate, Mannheim University library would need between 6 and 12 people to process all literature of social sciences bought in 2011 by Mannheim University Library.
Conclusion
While writing this blog post, I was struck by how much I mentioned linked data, in particular Wikidata. Given that Wikidata has rose dramatically to the 5th most important source of linked data in 2018, compared to 15th in 2015 it's no surprise I mention Wikidata a lot.
But why is linked data suddenly on my radar?
To be honest, I have always found linked data an almost impenetrable subject for years. A lot of it seemed to be specialists arguing over the most obscure points (e.g. ontologies, data modelling) and a lot of the demos I've seen were simply not exciting (displaying a link from the catalog to viaf) or did not seem to do much beyond what we already can do without linked data.
But given that libraries are now moving towards inside-out activitites there is a natural focus on enhancing discovery which linked data is meant to help facilitate. This is particularly true as academic libraries are now trying to collect and curate more than the traditional outputs of book + proving access to articles and the way to do it is unclear. Could linked data be the solution for enhancing discovery?
Is linked data moving up the slope of enlightenment?
In particular linked data in Scholarly Communications seems to be a very promising area, thanks to the efforts of Crossref who are building a scholarly infrastructure around relationships between different entities, add in open citiations becoming increasingly available in linked data formats, means that one can see many interesting applications. For instance one could detect if a paper one was thinking of citing was retracted as a feature in a reference manager. Or if citations were coded properly, one could find cites of a paper that agreed with the paper and those that did not.
The astute reader of course cannot help but notice a lot of the work I mention above require quite different skill sets from what many existing librarians already have. This is going to be a challenge academic library leaders are going to face increasingly - how to plan for transition .