web analytics

Universal search a core Glow service #EDUScotICT

Discovery of learning assets through search was an original requirement of Glow.  This should continue to be a requirement and furthermore, it is essential that we learn from the experience of Glow 1 to provide search in the most efficient and useful way.  I certainly do not want to dwell on the past but I will start this post by describing the original Glow search service which was not perfect

The Context
John Connell has often told the story of how when interviewing teachers, council officers and Directors of Education early in the SSDN project  (pre – Glow ) were asked what they wanted from Glow (looking forward ) – a common response was “Stuff………”.  It would be good if John could offer a deeper insight into this matter as there is underlying message!
There is no doubt that stuff was made available in and throughout the Glow Platform – but were teachers able to find the content which had been  uploaded – if they were not directed to it.  Sadly the original search facility in Glow as not capable of returning useful results to every search conducted – this was the case for a number of reasons.

Original Setup of Search in Glow
Glow is made up of a number of services each we have their own repositories.  The portal (MS SharePoint ) and the Virtual Learning Environment (Glow Learn – based on RM Kaleidos) to name only two examples.  There was a search capability on the portal page which could search only the portal repository and there was a much more “powerful” search tool in the Glow Learn which could search both the Glow Learn repository and some external sources (the search of third party or external repositories is sometimes referred to as Federated Search).

Search Scope
There were two search ‘engines’ available through the Glow Portal and Glow Learn as follows

  • Glow Portal Search – scope could return results referencing content which was stored in the portal (MS Sharepoint) repository.
  • Glow Learn Search – which could return results based on content stored in the Glow Learn repository and in addition links to some external repositories including the Learning and Teaching Scotland website and SCRAN.  The later was initially achieved using SRW/SRU (see here for some further information on this search standard http://techessence.info/node/48)

So there were two different places in the Glow environment where a user could access search which would return two independent results to the same query – this was not helpful.

Meta Tagging

In Glow learn it was possible to tag assets in a more granular manner.  Taxonomies were written so that teachers could tag learning assets according to the curriculum area to which the resource had relevance.  This opened up the possibility that a teacher looking for “Stuff………..” which was relevant to a particular learning outcome in the 5-14 curriculum (for example) would be able to search against a specific learning outcome and receive back a list of results which were relevant etc.  The matter of teachers having to Meta Tag resources that they were publishing did raise some concerns at the time and it was agreed that a minimum requirement would be that on upload, a meaningful title needed to be added to the asset.

Have things progressed?
Thinks have indeed progressed.  Later in the Glow project it was possible for the contractor to implement Open Search using the MS Search Server 2008.  This was applied to the Portal Search engine made possible a number of new and improved features as follows:

  • Users were able to have results returned based on their account permissions to access content.
  • Users could limit search results to include particular content types from the following list this Site, Documents, Spreadsheet, Videos, Acrobat, Images etc ( this list was configurable)
  • It was possible to use Open Search to include results from other Open Search enabled repositories.  (Currently, this includes ‘Published Glow Groups’, The Daily What News and the Learning and teaching Scotland WebSite)
  • The list of open search possibilities amounted to about many other sources such as bbc.co.uk, Flickr, YouTube, bing, Google safe search etc…….  but these have not been included to date.
  • Search could include results from Glow Blogs and Glow Wikis but this has to date not been implemented.  Ideally Glow blogs and Wikis should include a search API which supports identity if this is not currently available.

At this point it was still not possible to conduct search across both Glow Learn and Portal search engines.

So what are the requirements for Glow Future in this regard?
Whatever the future platform is for Glow I believe that the following needs to be openly discussed and a fully detailed set of requirements published.   Glow feature requirement should not necessarily be limited to the features which are currently generally available on internet based services.

I offer the following definition which is referred to below

A Core Glow Service is defined as a service which is accessible directly through the Glow Authentication ( or Single Sign On ) service.

I offer for the following suggestions for further consideration:

  • Glow should feature a dedicated search service which is accessible form the main portal but my also be presented to the user in other parts of the system (Glow Learn, Glow Collaboration tools etc…..) perhaps  this should be a standard tool bar type feature which is always available after login.
  • Search should be universal to the Glow Platform – this means that the underlying search engine should have visibility of and teh capability to search any and all repositories which are connected to Core Glow Services.  This should be considered as a base level requirement.
  • When searching for content on core repositories, users should have the capability to filter search results based on the asset ownership – so the user can see only assets they own, or assets owned but some other user where they have appropriate read permissions.
  • Search should only return to the user results which link to assets to which the current user has ( a minimum of) read access based on their credentials etc. so the target service must take the user identity as a search parameter.  This must apply to the results from core repositories and should also apply to result from federated search where user identity is relevant (this should include Google Apps if deployed, for example )
  • Federated search should also continue to be both possible and supported – this may require the continuation of either or both of SRU/SRW and Open Search.
  • Where possible Federated search results should be filtered based on the users credentials.
  • Having conducted a successful search a user should be able to refine the search – to for example find more resource published by and owned by the same author.

 

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.