Tuesday, November 02, 2010

Push Library search into Google : licenced database results in major search engines for "discovery"

It is not hard to stimulate people to use Google search. In medical education and practice we encounter a very large group using Google for many tasks: quick reference, known item search, citation checking, discovery, translation, spelling checks, address look-up, images, maps etc
Their motivation: "We believe everything we need right now is in there and its easy to find it".
Google Scholar even tops this up with the focus on "scientific" literature and a sort of citation scores.
It is hard to get people to use various licenced databases acquired by the library. In Google Scholar we push our linkresolver (sfx). This will appears automatically on any computer with our ip-ranges.
For off-campus users can set our library link resolver manually via Preferences.
But for search resuls in major search engines like Google, Bing and Yahoo there are no serious options to do this ... as far as I know of now. But this maybe a lack in my knowledge. I hope somebody will point me to any relevant info if it's out there somewhere.
I also would like to know any info on the projects that focus on showing link resolvers inside the normal Google.
I know Google Enterprise Search and Google Search Applicance GSA can offer federated search also for licensed e-content, but -again- I have no current knowledge of major academic institutions have adopted this technology. If they are out there, please tell me!
This short introduction showed that Google (or any major search engine) perhaps could become a preferred target to present your library e-content, via your link resolver or otherwise ...
This "otherwise" could mean showing direct live results from selected licensed databases NEXT to search results in Google, Bing or Yahoo.
This is exactly what WebMynd can do, a browser plugin/add-on for Firefox, Chrome and Internet Explorer.
I bumped into WebMynd via Aaron Tay (and his blog Musing about librarianship). Please read his interesting post: Adding your library catalogue results next to Google results using WebMynd
With help from WebMynd staff, I tried to set up 5 main resources for our medical library as a test. My goal and interest is to see how library resources can be setup to appear in these browsers to improve discovery.
  • PubMed : with our linkresolver RUGLinks (SFX)
  • UpToDate: awkward to test because of its special Accept Licence page (via proxy)
  • PiCarta: Dutch central Catalogue which includes our local holdings (via Proxy)
  • Scopus: one of the biggest abstract and citation database by Elsevier (via proxy)
  • AccessMedicine : large medical e-book collection, fulltext searchable (via proxy)
This features are now available by WebMynd
  1. WebMynd Install for users (just discover the web, for people not planning to create their own Sidebar)
  2. For simple Sidebar setup you can only fill in the website url. I assume they use Opensearch and some manual tweaking for this. You will get an email when the sidebar is ready
  3. Advanced Mode: WebMynd currently offers the option to create your own customized sidebars with the use of : 
    • a web query string and
    • CSS Selectors , based on JQuery (more info)
      (setting this up is a little simular to the use of some Feedscraping tools, but the testing feels a bit wobbly. Maybe a more extensive guide with certain often used elements and more examples is helpful)
    • the addition of a favicon is not required, but looks cool when available. I use GetFavicon to find the right urls to the favicons.
Here is the test Install Page for this Sidebar with the above mentioned search engines. (Remember you need our proxy to see results)
This is how it looks in Chrome (after proxy-login)
This first beta library Sidebar with Proxy functionality works reasonably well. At first glance it is impressive to see results from those licensed resources.
From outside university campus you will see the word LOGIN inside the individual items that indeed need authentication. In this stage WebMynd only offers login after clicking inside that item and filling in the account or proxy details. These will stick as long as the browser session takes, or -in case of some resources- as long as the "session-id" of the original resources allows to keep it open.
I hope they will consider making a Settings Tab inside the WebMynd Sidebar when users can securely fill in several account details, like Google, Facebook, Linkedin, Library proxy etc. Maybe they can even think of an OpenID to use.
So far the guys at WebMynd's were very response and acted quickly to various suggestions. They appear to be taking this further seriously, possibly with a special focus for libraries.
Right now I am testing this Sidebar within Chrome, FF and IE.
Some early observations:
  • the proxy login needs to be activated inside one box. This opens up a new windows, but the previous window does NOT get a refresh automatically to show the hits. I had to do this manually.
  • the CSS selectors need to be tweaked to present the results consistenly in the same way for every resouce.. I would advice them to try to show scientific literature always in max 3 lines: title info, authors and citation details. The tweaking is not made easy in the current layout and design. Updates are already announced.
  • The above calls for more sophisticated setup via API's when available. Currently they only do this on demand. It is not yet added to the Content Concierge.
  • URL's work fine and show our link resolver images.
  • UpToDate is, with its two-step authentication, still awkward.
  • When your search term scores no hits, you get no mention of this, other than a simple no show. Strangly Scopus shows its login screen, instead of not showing at all. Maybe it is more logical to show it on top with "no hits". You can do this manually by selecting the resouces again in the Personal Tab.

  • Inside Firefox it works simular as in Chrome. Currently the IE install is giving me a hard time. Maybe one of the numerous addd-ons I am using, is conflicting with WebMynd. We are trying to figure out this.
  • Also the layout in IE doesn't looks not as smooth as the others. Some texts overlap.
I am enthusiastic about this sidebar search and had great fun exploring the additional resouces already available in the default sidebar. This way I discovered great materials out there on the web for various topics.
Adding other web resources or channels was very easy. Here are some of the included tools.
WebMynd Sidebars work with Google Search, but also for Bing & Yahoo. When questioned WebMynd Support states that it is not really difficult to activate other search engines as well, but that they are doing research into which ones to take on next. I suggested Google Scholar to get into the audience of scientists, researchers and people who are interested in scientific literature.
Basically a library could create, like tailor-made searchboxes or subject portals, Sidebars with focussed databases for patrons, departments etc.
How do you see the use of WebMynd for subject and publication discovery? 
Enhanced by Zemanta

1 comment:

Aaron Tay said...

Hey Guus.

Great to see you took the next logical step.

I was trying myself with Scopus with the help of webmynd, it worked fine as the "pattern" is the same.

A minor concern is that the sorting order of Scopus result is not by relevance for me, but that's a small matter.

I wonder though if library vendors might object to Webmynd scraping the results like this? Might be against TOS for some of them?

I was thinking though for some databases like Sciencedirect,JSTOR & Project muse, where results can be shown without login, if it might be better to just scrape the results from the non-proxied page but the result link to be modified with the proxy link? (This is similar to how Libguides handles RSS feeds i think)

One advantage would be the results could be seen even before login.

Of course, to truly cater to library users, would be even better if users could just put in the ezproxy string and it would work for some of the popular databases.

Just some thoughts from someone who isn't very techie..