Thursday 4 July 2019

A comparison of 6 Access Broker browser extensions Lean Library, Kopernio, Anywhere Access, Libkey Nomad & more

Source: http://musingsaboutlibrarianship.blogspot.com/2019/07/a-comparison-of-6-access-broker-browser.html

Given that Browzine has launched Libkey Nomad, their take on browser extensions that help direct users to full text available - both open access versions and those available via institutional edition, it is a very good time to do a technical comparison of the alternatives available on the market. (See also earlier comparison in May 2019 that gives more historical context )



Lazy Scholar, AnyAccess, Kopernio, Google Scholar button, Lean library, Libkey Nomad (From top left, clock wise)

I compare the following (in rough order of age)
You might be wondering why I am excluding Unpaywall, Open Access Button or the newly announced CORE Discovery browser extensions. The reason is they focus only on free to read articles and do not direct users to institutional access versions which is the topc of interest here.
The important thing to note is that Unpaywall, CORE discovery, Open Access Button etc also exist as APIs as well as extensions and are often called by the other extensions I am comparing to check for free copies (more on this later).
A warning : Information on the comparison table is based on my understanding talking to the different vendors, attending demos and reading their website and I do not guarantee 100% accuracy. In some cases, I have personal experience with the extensions (Lazy Scholar, Google Scholar Button, Kopernio), while others are only via demos by vendors, talking or reading websites, so do not take everything here as gospel and do your own homework!
The comparison table is available as a Google Docs , feel free to comment if you think there are inaccuracies or improvements. 
I've done a historical overview of these services and extensions just 2 months ago, so I will skip that and focus on specific comparisons of areas of interest to librarians thinking of implementing them.

Business model compared


Product Vendor Business Model Available since (estimate)
Lazy Scholar Colby Vorland Free 2014
Google Scholar button Google Free 2015
Lean Library Lean Library (Sage) Subscription 2016
Kopernio Kopernio (Clarivate) Freemium* 2017
Anywhere Access Digital Science Subscription 2018
Libkey Nomad Third Iron Subscription 2019


Of the 6, Google Scholar button and Lazy Scholar are free. While the rest are available via subscription. Kopernio is also currently free but has a premium version.

First let's talk about the most popular browser extension with 2 million downloads - the free Google Scholar button. I once commented on the irony that Google Scholar was better at supporting discovery of open access items than our discovery services. While library discovery services are doing a bit better now, particularly with the intergration of OA finding services like Unpaywall, it is still undeniable that Google Scholar (and Google) is the major source of traffic to open repositories and probably still the best at finding free to read articles.



Also see analysis by Petr Knoth - showing a combination of various extensions (excludng Google Scholar) outperforms Unpaywall (or any single extension) by a third at least, showing such tools still can improve in recall.

Some of it is no doubt due to the fact that Google Scholar indexes ResearchGate and Academia.edu, while Unpaywall doesn't (due to doubts about legality of papers there) but this might not be the full story. For example, if a researcher puts up a copy on his personal homepage and not in a repository it is unlikely Unpaywall or others can find it, while the all seeing (feels like it) , Google Bot will spot it easily.

Of course, the main issue with Google Scholar is that there is no API for others to use, but this is no barrier to Google themselves and they have the highly popular Google Scholar button by the official Google team.

Their Google Scholar button leverages Google Scholar and library link programme (which most academic libraries participate in) and their already mentioned OA finding capabilites to allow users to access Google Scholar's excellent delivery experience even when they are not on the Google Scholar search result page.
Google Scholar button leverages the Findit@xxx links most academic libraries have set up in GS

Google Scholar button extension surfaces link to library full text (same as link in Google Scholar)




The other free extension is Lazy Scholar and the first version of it actually uses the same idea to find full text by scraping the links. The astonishing thing is, it in fact predates Google Scholar button and amazingly it is the only one in the list that is supported not by a company but is a hobby project by a single person - Colby Vorland. It is in fact the oldest browser extension in this list, started circa 2014.
Kopernio another early pioneer was founded by  Jan Reichelt and Ben Kaube, the co-founders of Mendeley and Newsflow. It was acquired by Clarivate in April 2018, and their business model is somewhat hard to classify. Currently it is free for use, but has a premium edition. I am tempted to classify it has freemium because of that, but as at time of writing, you get premium by recommending a friend to use it which is a low bar. 

Anywhere Access is by Digital Science was announced in July 2018, and my understanding of it is the most minimal. The webinar showcased a browser extension, and it can be intergrated with Dimensions and ReadCube. This is available via subscription
Lean Library browser was a spin off from Utrecht University Library's UU Easy Access (circa 2015) by  Johan Tilstra and was eventually acquired by Sage in September 2018. Like Kopernio, Anywhere Access and Libkey Nomad it is available via subscription.
Libkey Nomad by Third Iron (Browzine) is the latest to join the party in June 2019. It is also available only via subscription but if you are already an existing customers of Browzine you automatically get both Browzine and the Libkey suite until your contract run out (please check with your Browzine rep).

If you are curious these are how many downloads each browser extension has from the Chrome store, here are the statistics as of June 30, 2019. Obviously this doesn't mean a lot as some extensions like Kopernio had a big headstart, and some extensions are free , others are not.

Downloads of each extensions from Chrome Store as of June 30, 2019 
They all also have different strategic goals and plans. Lean Library has probably been quickest to sign up academic libraries (thanks to librarian friendly features like Libassist) and has signed up Universities like Havard, Stanford University pretty early, some even prior to acquisition by Sage.

Kopernio seems to be pursuing a different strategy, and has been offered free so far to build up a user base  (shades of Mendeley - whom Jan Reichelt was founder of as well ) , this has allowed them to ramp up quickly. At the time of writing , even though premium account exists, it is unlocked by recommending a friend, and I'm unaware of any subscription fee to unlock it at individual or institution level. Given it's acquisition by Clarivate one can imagine a possible bundle with Clarivate Web of Science etc though this is just speculation.

The rest are too new at this point to really tell what their strategy will be though mostly likely they will build on their respectively base i.e. Anyway Access will appeal to Digital Science Dimensions customers, while Libkey Nomad has a build-in base with Browzine customers.

Either way this is still a very new market, most users do not yet use anything.



My view  :  I have been using Google Scholar extension since it first began, and it has the usual Google ethos of making things has  simple to use as possible so if all you want is ease access to full text or if you are brand new to this type of products, you should definitely give it a try and consider whether it is good enough for your purposes.  A slight drawback is this extension unlike most of the others usually does not automatically check for full text whenever you visit a page, and you must remember to click the button.

Colby Vorland's Lazy Scholar which I reviewed way back in 2014 is a more complicated extension with a ton of features and definitely worth a look due to its variety of functions, but I can imagine librarians may be a bit worried on relying on an extension that is supported by a simple persn

Of course, Librarians and institutions might want to implement services that have a clear business model (who knows when Google will graveyard the extension) and provide support and offer features beyond simply redirection to full text such as ILL/DDS and other library support services. Some analytics would be nice too. 

The remaining four are all backed by established companies, so it depends on your views of the parent companies. 

These tools likely will or have the potential to link up with sister products and services , for example Kopernio with  Clarivate's EndNote (X9.2) /Web of Science/Publons, AnywhereAccess with Dimensions/ReadCube, which will obviously be a consideration.

I can imagine bundled packages might also start appearing for some of the acquired tools. 

The degree of intergration will of course vary on the strategic priorities of each product e.g. I currently do not see any moves to intergrate Lean Library with Sage stable products, though that might change at any time of course.

Access to open access/free to read copies

Product Open Access Sources
Lazy Scholar Google Scholar (scrape), Dissem.in, EuropePMC, Pubmed, PMC
Google Scholar button Google Scholar of course
Lean Library Unpaywall + others
Kopernio Unpaywall + own sources/crawler?
Anywhere Access ?
Libkey Nomad Fully Gold OA journal only


It's generally difficult to tell what sources these browser extensions are using to find free full text.
The clearest one is Google Scholar extension which of course is using Google Scholar itself. Lazy Scholar extensions which began life scraping Google Scholar search page in the background also does the same.

Most of the others are likely to be using a mix of resources (perhaps via API calls or preharvested) such as 
  • Unpaywall, Open Access button, PMC, EuroPMC, BASE, CORE, Crossref, Dissem etc
My sense is most of them don't aim to push the envelop on this and rely on existing OA finding infrasctructure as this isn't the most important function.
In fact, even those "pure" OA only finding services often use the same sources and many such as Open Access button and the new Core discovery pull from Unpaywall etc.



My view  :  This probably isn't a very important point of comparison, since such tools can pretty easily reach parity by using the OA finding infrastructure that is quickly maturing and you probably want to focus on it's ability to find subscribed versions.

 That said, there might be differences in OA finding capability. For example, as noted above Google  Scholar button and Lazy Scholar which relys on it is likely to find more free to read articles. 

 Also as of writing Third Iron has confirmed with me that for Libkey Nomad their OA finding capabilities are basically same as Browzine - finding articles in Gold Open Access Journals (probably not hybrid?) , if so they will be finding a bit less OA material at least for now than the others (which rely on Unpaywall or similar methods that index at a article level) and are capable of finding Green OA etc.

 So I suspect in terms of recall of free papers  -   Google Scholar/Lazy Scholar >  Kopernio > Others > Libkey Nomad but testing is needed.

Some thoughts ; What is it the false positive and false negative rates of these tools?  What sources are they pulling from (are they legal?) ? What versions (published version, accepted version, submitted version) are offered (can they be configured)? Institutional repositories of course should check their contents are discoverable just in case.


Requires Submission of library holdings / Method of detecting entitlements

Product Uses your library link resolver? How it detects your entitlements and redirects Requires librbary submission of holdings?
Lazy Scholar Yes, Scrapes Google Scholar link (secondary method) It queries various publishers with your user credentials (ezproxy) and checks for pdf (primary method) + Scrapes GS Library link resolver link (secondary method) No, if you already setup Google Scholar Library Links
Google Scholar button Yes Library link resolver link + holdings you give Google Scholar No, if you already setup Google Scholar Library Links
Lean Library Yes Library link resolver link + holdings you give Lean library Yes
Kopernio No It queries various publishers with your user credentials and checks for pdf No
Anywhere Access No? ? Yes
Libkey Nomad Only as fall back Libkey infrastructure (same as browzine) + holdings you give browzine No, if you already setup holdings with Browzine
One of the more interesting questions about browser extensions like Kopernio, Lean Library etc compared to pure OA finding only ones like Unpaywall is how does the extension know what you can access, display the right icon and then redirect you to it?

This is actually a two part question. First the extension needs to know what you can access and display the right popup/icon and then redirect to the PDF using either it's own redirection service/resolver or use the library's link resolver. But let's focus on the first question, the answer seemed obvious, the library/institution has to give the holdings of what they own to these services.

Electronic resource librarians reading this will probably go, yet another workflow to setup but in fact, you might not actually have to do this!

The most obvious reason is that the browser extension already has access to your holdings in some way.

For example, if you are already a Browzine customer, you obviously are already sending your holdings to ThirdIron so their Libkey Nomad extension can use it.

Similarly you probably already send your link resolver holdings to Google for the library link programe in Google Scholar, so the Google Scholar button  extension and Lazy Scholar extensions (which scrapes the GS result page work automatically for your institution without you lifting a finger.

That said, it might be possible to avoid even that.

Kopernio and Lazy Scholar (primary method), do not require your holdings at all. This sounds like magic but it basically queries the publishers on your behalf using your stored credentials to see if a PDF is available. This may or may not be acceptable in your IT policies.


Lazy Scholar configuration screen - "deep linking"  using Lazy Scholar will use your institutional access to check for full texts

Here's how the author of Lazy Scholar describes it - "Lazy Scholar stores the encrypted institutional login info in the browser cache once you authenticate through the extension, similar to Kopernio's solution. So when the proxy session expires it uses this to start a new session in the background and look for the full text." (Colby Vorland)

This is how Kopernio describes it on "How does Kopernio store and protect institutional credentials?"



Here's how it looks the first time you use Kopernio, where it offers to store your credentials. Note it says you "only have to do this once"


Kopernio offering to store your credentials
Of course Kopernio and Lazy Scholar are careful to state that the credentials are "never transmitted to the Kopernio servers. Credentials do not leave your browser other than to be sent to your Institution’s Authentication Page."

For Kopernio you can also choose not to store credentials but whether even with this option, your IT department may or may not have issues.

Kopernio setting to turn off persistent credentials 



Lean library like the other extensions requires holdings for their "Library Alternative" module to work (KBART, or direct via various API supported like Alma, SFX etc  ) but has an additional quirk that you need to provide access to your ezproxy config files for domains to be proxied. for an additional function called "Library Access".

This unique function recognizes domains that can be proxied and when the user lands on those domains it will offer to automatically do it for users (this is somewhat similar to Zotero proxy redirection).

This can be useful for database domains that don't necessarily have the usual journal full text e.g. Scopus, Lexis-Nexis

My view  : Kopernio type methods need to be reviewed by IT security for compliance. Otherwise it seems to me this is a one time setup cost mostly, so shouldnt be a major consideration unless it affects accuracy and reliability of redirection (see next section).

Support of publisher and aggregator platforms, 1 Click PDF

Product Aggregator support 1-Click to PDF
Lazy Scholar As per library link resolver (secondary method) Yes (primary method)
Google Scholar button As per library link resolver No for those via librarylink resolver
Lean Library As per library link resolver No
Kopernio No, not with Proquest or Ebsco Yes
Anywhere Access ? Yes
Libkey Nomad Yes, list Yes
The meat of the extension is how well it directs users to papers they have institutional access.  While libraries have a lot of experience with library link resolvers, and if all the extension does is pass on the metadata of the article they detect to your library link resolver, things are a bit more visible and  to the librarian and you retain some control over this aspect.
However,  many of these extensions are not using that but rather use their own resolver/redirect service (e.g. Kopernio, Lazy Scholar (Primary method), Libkey Nomad).

If so, this is a service that is transparent to you and you do not control directly, so you need to test it carefully, in particular for full text available only via aggregators rather than publishers.

Library Link resolvers in particularly OpenURL resolvers are of course designed to solve this so called "appropriate copy problem" and if the browser extension uses  your link resolver, you probably have this problem licked (as much as possible anyway given that link resolver link stability hovers around 80% despite efforts by NISO etc. ).

So for example if you are using Google Scholar extension or Lazy Scholar (secondary method), you have aggregators covered since they use your link resolver directly.

However if your browser extension does the redirection on their end, you should definitely check if they cover aggregators (ask for a list).

Kopernio for example currently does not support aggregators even the major ones like Proquest and Ebscohost.

What about Libkey Nomad that uses the same mechanism (now called Libkey) as Browzine? If you remember the early days of Browzine they had very little support for aggregators, but they have improved since then and I spot Ebsco and Proquest in their list now.




Still i have a sneaking suspicion the typical library link resolver such as SFX, 360link, Alma Uresolver is typically capable of covering more sources though at the cost of less link reliability, though I would add today's library link resolvers go beyond just OpenURL but also doi/pid based linking , IEDL (Index-Enhanced Direct Linking) etc.

You should also check, if these proprietary redirection services handle only publications with an doi (eg. Libkey Nomad only works with publications with a identifer like PMID or DOI ), or if they can work with anything with sufficient metadata like a typical library link resolver.

The one click to PDF advantage?

Many of these extensions like Kopernio, Anywhere Access and Libkey Nomad have in their marketing literature a phrase like "1 click to pdf". This is touted as a advantage over traditional link resolvers.
Often it is explained by drawing a parallel to how in Sci-hub, you enter a doi and you get a PDF directly to download.


In comparison, when you click on a link that is mediated via the library link resolver you might see the following jumps before reaching the PDF

  • The link resolver menu/service page which you might be given multiple choices of which database/source to use to get the full text , so you click on one
  • The article landing page on the publisher site , so you hunt for the download button
To be fair, many link resolvers like 360link, SFX, Alma Uresolver can be setup to skip the link resolver menu page if a condition is met (e.g Only one possible full text source AND no access note for user).

Also many of this tools that boast this feature do not guarantee 1 click PDF to all sources (see Browzine's list)
Others might worry that jumping to a PDF direct will be problematic if the link is broken and the user will be "lost" navigationally speaking, and some have tried to solve with Iframes (e.g. 360link, Kopernio)


Kopernio sends you direct to PDF but with a Iframe

I am unsure if other browser extensions that do not market 1-click PDF can do this has well so this is worth testing for if you think is important. In fact, one wonders if library link resolvers can do this too.

Using Library Link resolver as a fall-back?

While the newer redirect/resolver services tend to be more stable in linking and might provide better experience, I suspect they are more limited in coverage than traditional library link resolvers.

As such it might be some of these services have a fall back to your link resolver if their own system is unable to resolve the full text. In a way this gives you best of both worlds, but it may be in those cases the Extension will show unavailable and only clicking in through to your library link resolver will they realize it is available.


My view : This is a big one. Besides user experience, you probably want to spend most of your time testing how good the browser extension is directing your users to content. You might get away with less testing if it is more or less just sending metadata to your link resolver directly (e.g. Google Scholar button) but even then the extension capabilities might differ on how good it is at parsing information on the page you are on. 

 I would focus on testing by sources particularly big aggregators to see if the extension is able to direct your users there. The usual stuff about false positives and negatives (including link reliability) applies of course.

Library support services - ILL/DDS etc

Product ILL/DDS Support
Lazy Scholar No
Google Scholar button No
Lean Library Yes
Kopernio No
Anywhere Access Yes, Readcube intergration?
Libkey Nomad Yes


A very natural thing for these browser extensions is to handle cases where there is no access (institutional or free) and offer ILL/DDS service.

This can be handled in a variety of ways such as just sending you off to the library link resolver for ILL/DDS service (e.g. Lean Library, Libkey Nomad) or something more intergrated such as via ReadCube

My view : I expect this to be a standard feature eventually

Analytics and Privacy

This consists of two parts. Does the extension provide data that is useful for you as a librarian, yet at the same time limit the compromise of user privacy?

As I blogged, some of these extensions such as Kopernio, has Support of COUNTER R5 and distributed usage logging( DUL).
Support of COUNTER probably helps you compare apples with apples to see how much of downloads are driven by the extension against total downloads. But some might not like the idea of submitting the stats to publishers for aggregation due to DUL.

Because the extension looks for both subscribed version and free to read versions, such extensions can give you an idea of how much the user is requesting that is already free (and also where, and what version). This is very useful to aid for negotiations of course.

Anywhere Access lists "Knowledge base error monitoring" which seems to be very useful obviously. 
My view : Developong area. Room for librarians to suggest statistics that are useful for them. Some standards here might be nice so we can compare across services.


Things that caught my eye

I'm not going through the whole table, and I've focused on the more technical bits, but an assessment should obviously cover user friendliness (e.g. does the extension popup automatically when it detects free full text? Is it annoying? eyecatching enough? Responsive?), ease of implementation and roll-out (can you mass deploy across all university computers?), browsers supported, security , marketing plan etc.

Also each service has also unique features - off the top of the head, these caught my eye
  • Library assist feature (Lean Library) - Popups custom messages when users visit domains
  • Overlay links (Libkey Nomad, Koperino, Lean Library) - overlays links in Google Scholar (Kopernio, Lean library), Wikipedia, Pubmed (Libkey Nomad, Lean Library)
  • Intergration of PDF with Endnote, Dropbox  (Kopernio)
  • Install ILL/DDS via Readcube (Anywhere access)
  • Ebook access (Lean Library)
  • Knowledgebase error montioring (Anywhere access)

Kopernio overlaying links in Google Scholar

Conclusion

Access brokers browser extensions is in my view a interesting "Second coming" of Library 2.0 toolbars.  The first time around circa 2007 , it died a still birth. But this time around the commericals many backed by big publishers and vendors have thrown in their hats in the ring, will it do better this time?

Will libraries be able to market this such that it is installed and used by more than a vanishing small percentage of their user base?

It is still early days of course, other vendors might be throwing their hats into the ring soon. 
Posted by at

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)